A systematic review of approaches to assess fish health responses to anthropogenic threats in freshwater ecosystems

The mechanisms through which human modification of freshwater environments impacts individual fish health are diverse and complex. We present a systematic review of research spanning 50 years to identify impact pathways of freshwater human-induced threats and the health assessment methods that are sensitive to them.


Introduction
Freshwater biodiversity constitutes a valuable resource, providing important economic, scientific, aesthetic, cultural and recreational ecosystem services (Dudgeon et al., 2006).Despite this, global freshwater ecosystems are some of the most poorly managed, with anthropogenic disturbances occurring simultaneously with natural environmental variability.While anthropogenic pressures on freshwater ecosystems continue to mount globally, there are proportionally few studies that investigate the impacts of these pressures on individual fish health (Bergman et al., 2019).Recent studies and reviews (e.g. Watson et al., 2020;Brosset et al., 2021) have described how particular stressors can impact physiological parameters with consequences for fitness and overall population dynamics.Here, we develop a conceptual framework that links stressors, fish health, fitness and population dynamics (Fig. 1).Research into the link between anthropogenic threats and individual fish health (Link A, Fig. 1) is limited, when compared to the body of research related to understanding the consequences of fish health for individual fitness, population dynamics and higher levels of ecological organization (assemblages and communities) and the study of their interrelationship (links B and C, Fig. 1) (Birk et al., 2020;Brosset et al., 2021;Pinna et al., 2023).
Despite this, the importance of individual health is increasingly being recognized in conservation management, and the discipline of conservation physiology has emerged as a new science (Cooke et al., 2022).Conservation physiology aims to use various physiological health components to understand how both natural and anthropogenic disturbances (or a combination of both) can translate into population level changes (Cooke et al., 2013).In this context, physiological health components refer to cellular, organ and organismal function, including aspects such as behaviour.The potential utility of these health components as diagnostic tools is evident for several reasons.First, physiological fish health components can be highly sensitive to anthropogenic disturbances, and changes in health components are measurable well in advance of demographic changes, potentially affording sufficient time to implement threat mitigation strategies (Ellis et al., 2012).Second, physiological indicators have a direct link to drivers of population dynamics such as individual fitness (Bergman et al., 2019).Finally, the assessment of physiological health components can provide a robust and mechanistic understanding of the impact of anthropogenic stressors on an environment, as opposed to demographic indicators such as abundance or biomass (Horodysky et al., 2015).Despite the apparent benefits of this discipline, the application of physiological biomarkers as management tools has only recently been applied to conservation challenges (Cooke and O'Connor, 2010;Madliger et al., 2016;Bergman et al., 2019).
In this review, we aim to provide an understanding of the initial pathway through which anthropogenic threats in freshwater ecosystems impact fish health, and synthesize the ways in which fish health components impacted by anthropogenic threats have been assessed as health indicators.We analyse a corpus of papers that assess the impact of multiple anthropogenic threats (e.g.water infrastructure, climate change, overexploitation, invasive species and landuse changes) on freshwater fish health.Using a quantitative systematic approach, we also quantify spatial and temporal trends within our corpus.

What is fish health?
Fish health comprises multiple physiological and behavioural components, each of which operates on different biological and temporal scales.Selected components of fish health are responsible for maintaining homeostasis-the maintenance of molecular, cellular and physiological function within a 'normal' range (Segner et al., 2012;Yıldız and Seçer, 2017;Shiry et al., 2023).The maintenance of homeostasis in turn enables other components of fish health to operate, such as growth to maintain condition, reproduction and locomotion (McEwen, 2000;Segner et al., 2012;Lloret et al., 2013;Sokolova, 2013;Schinegger et al., 2016;Yıldız and Seçer, 2017;Balasch and Tort, 2019).
One way of summarizing components of fish health and how they interact with the environment is the categorization of responses to stressors (Wendelaar Bonga, 1997).Stress responses can be categorized into three main groups: primary stress responses, secondary stress responses and tertiary stress responses (Mazeaud et al., 1977;Wedemeyer and McLeay, 1981;Barton et al., 2002;Portz et al., 2006;Segner et al., 2012) (Fig. 1).Each level of stress response is comprised of different components that can be measured as a health indicator to infer overall fish health.Summarizing fish health in such a way helps in understanding that different components are sensitive to different types of stressors and respond on different time scales.Primary stress responses involve changes in neuroendocrine function, resulting in the release of hormones such as cortisol (Pottinger, 2008;Sheriff et al., 2011).These hormones are released into the blood, where they can be measured to assess responses to stressors and initiate other physiological responses that make up the secondary stress response (Barton, 2002).Secondary stress responses primarily include changes in metabolic function, changes in immune system function, changes in haematological features and cellular changes such as increased heat shock proteins (HSPs) (Barton et al., 2002), each of which are measurable as health indicators.Secondary stress responses may temporarily tolerate acute stressors and allow a return to homeostasis; however, if the stressor is severe or occurs for a period of time outside of regulatory capacity, primary and secondary stress responses may culminate in tertiary stress responses (Sopinka et al., 2016).These stress responses are defined as whole-animal changes and include morphological changes such as decreased condition (Segner et al., 2012).Condition is another health component used as a surrogate fish health indicator (Brosset et al., 2023).There are also other methods  'Stressor' refers to anthropogenic threats and their resulting environmental alterations and impact mechanisms (described further on) that operate in freshwater environments.'Fish Health' refers to health components that make up the primary, secondary and tertiary stress responses.'Fitness' refers to parameters including but not limited to reproductive capacity, migratory capacity and survival.'Population dynamics' refers to factors such as recruitment, population resilience and biomass.The letters 'A', 'B' and 'C' highlight the links between these steps.
that can be used to assess fish health that do not fall within these definitions.For example, the presence of parasites does not fall into these categories, as they can occur irrespective of physiological processes and health status.Details of commonly measured stress responses are provided in Table 1.A more detailed description of stress responses is provided in Supplementary Material S1.

Linking anthropogenic threats to fish health
To employ physiological biomarkers as effective management tools, it is important to understand how anthropogenic threats influence and modulate stress responses within an ecologically relevant framework (Cooke and O'Connor, 2010).Anthropogenic threats illicit a broad range of stress responses in fish and the mechanisms by which these threats ultimately impact fish health can be unclear (Sokolova, 2013;Schinegger et al., 2016).Previous reviews have created hierarchical frameworks that step out these mechanisms.For example, Craig et al. (2017) define anthropogenic 'threats' as humancaused drivers of environmental change and environmental 'alterations' and ecological 'effects' as the environmental changes they produce and the ecological responses to those changes, respectively.Our review adopts a similar framework, however, the term 'impact mechanism' has been used instead of ecological 'effects' as our effects refer specifically to the mechanisms by which fish health can be directly impacted, and in some cases, these are not through changes in ecological function.We define unique combinations of anthropogenic threats, environmental alterations, impact mechanisms and health responses as an impact pathway.Figure 2 shows our impact pathway framework in use, with the anthropogenic threat 'Water Infrastructure' used as an example.
In our review, this framework is applied to the five major anthropogenic threats to freshwater ecosystems (Dudgeon, 2019).These anthropogenic threats are 'Water infrastructure', 'Land-use changes', 'Invasive species', 'Overexploitation' and 'Climate change'.Water infrastructure pertains to the construction of dams, weirs and other barriers, resulting in environmental alterations including flow alteration (i.e.changes to the magnitude or timing of river flows), habitat alteration (loss of connectivity or critical habitat), abstraction, river impoundment and changed water quality (Fig. 2).Land-use changes refers to agriculturalization, industrialization and urbanization.Pollution is often considered as a broader anthropogenic threat; however in this review, it has been included as a sub-category of land-use (Tahiru et al., 2020).Freshwater ecosystems are particularly vulnerable to pollution, with chemicals, heavy metals, microplastics and other particulate matter being of greatest concern (Bashir et al., 2020;Gokul et al., 2023).Other environmental alterations deriving from land-use changes include the removal of terrestrial vegetation and flow alteration.The introduction of invasive species into freshwater ecosystems is primarily a result of anthropogenic practices, either intentionally or unintentionally, leading to impact mechanisms such as increased competition, predation and the introduction of disease and habitat degradation.Overexploitation refers to the unsustainable use of freshwater resources, encompassing recreational and commercial overfishing and the impacts of handling on both target species (e.g.catch-and-release recreational fishing) and bycatch (e.g.commercial bycatch returned to the water).Climate change acts as a ubiquitous threat to freshwater ecosystems and has the potential to exacerbate the effect of the above anthropogenic disturbances.

Literature search and database development
The review protocol followed that outlined by PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) (Supplementary Fig. S1) (Page et al., 2021)    theses) form is included in Supplementary Sheet S1.While a systematic protocol was followed in the creation and categorization of our corpus, we refrain from asserting confidence in our control for validity and bias reduction, as we did not conduct critical appraisal of each paper to check for any biases in experimental design, data collection or analyses Refining the corpus further occurred as a two-stage process: Stage 1: screening and Stage 2: categorisation (Supplementary Fig. S1).The first stage involved rapid assessment of the title and abstract to check for relevancy, with all papers that measured fish health in some way retained.Papers that did not aim to measure fish health were excluded (e.g.papers that assessed fish palatability or the relation of fish health with the production of fishbased products for human consumption).Duplicate papers, grey literature, reviews and meeting abstracts were also removed during this stage, with a total of 2522 papers excluded from the corpus.During the second stage, the remaining 1875 papers were categorized to generate a dataset for the quantitative systematic literature review.A further 1514 papers were excluded as they did not assess responses in fish health to an anthropogenic disturbance, leaving a total corpus of 361 papers.Papers were then categorized by geographic location, study type (laboratory, field or a combination of both etc.), anthropogenic disturbance type (anthropogenic threats, environmental impacts and impact mechanisms, described above) and stress response.To categorize the anthropogenic disturbance type, the following definitions were used.Anthropogenic threats refer to the broad, basin scale threats that impact freshwater ecosystems.Environmental impacts refer to either the resulting changes in an environment that occur because of an anthropogenic threat (e.g.river impoundment as a result of water infrastructure) or a secondary threat as a result of an anthropogenic threat (e.g.fishing as a subgroup of over-exploitation).Impact mechanisms refer to the various mechanisms that have a direct impact on fish health components.For example, changed food availability can directly impact whole-body morphology, organosomatic indices or lipid and fat content.
Due to the large number of ways in which the primary and secondary stress responses can be measured, assessment methods belonging to these categories were grouped.Health components that belonged to the tertiary stress response, or health components considered to fall outside of primary, secondary or tertiary stress responses were individually categorized.Studies were also categorized by whether they were conducted in either an artificial setting, the field, a laboratory or a combination of field and laboratory settings.Laboratory studies included experiments on both living and dead fish.Field and laboratory studies were defined as studies that had both field and laboratory components.For example, the sampling of fish from different rivers where an attribute of the river (e.g.pollution) was predicted to have an impact on the result, but further laboratory analysis was required to obtain results.Field specific studies were those conducted entirely in the field, employing methods that did not require any further laboratory processing.Artificial studies consisted of studies either ex situ, or those conducted in situ in a separate enclosure.These studies were not classified as laboratory studies as they included the use of mesocosms or purpose-built enclosures where the physical size and shape of the tank were important to the study (e.g.simulating the presence of a fish screen or artificial sound/light).Further detail of the fish health categories used in this review is provided in Supplementary Table S1.The complete corpus and categorization criteria for this study are provided in Supplementary Sheet S2.

Quantitative summary
The dataset generated from paper categorization was used to investigate: (i) the geographical distribution and temporal variation of studies within the corpus, (ii) the frequency of different study types, (iii) the prevalence of different fish health assessment methods used to evaluate responses to anthropogenic stressors and (iv) the impact pathways of anthropogenic threats that results in changes to fish health components.To quantify impact pathways, the anthropogenic threat discussed in each paper was broken down into three hierarchical categories: (i) the type of anthropogenic threat, (ii) environmental alterations (the way in which these anthropogenic threats alter the environment) and (iii) impact mechanisms, the mechanism by which environmental impacts cause changes in fish health indicators.Anthropogenic threat categories, environmental impacts and impact mechanisms were adopted from multiple reviews of anthropogenic threats in freshwater ecosystems (e.g.Dudgeon et al., 2006;Dudgeon, 2019) and expert knowledge.These were then linked with the type of fish health indicator each study employed to understand the mechanism by which anthropogenic threats changed components of fish health.This information was presented as an alluvial diagram, a type of flow chart that can be used to represent nested hierarchical structures.For this visualisation, we utilized the 'ggalluvial' package (Brunson and Read, 2023) in R Studio (R studio version 4.3.0).

Results
Papers that assessed the impact of anthropogenic disturbances on fish health in freshwater systems were derived from 49 countries.Of these countries, the most studies were undertaken in the United States and Canada (combined) (n = 93) (Fig. 3).Of the remaining 312 studies, 105 were from Asia, 73 were from Europe, 56 were from South America, 17 were from Africa and 16 were from Oceania (Australia and New Zealand).
We found that studies assessing the links between anthropogenic threats in freshwater ecosystems and fish health components have increased through time from 1972 to 2021 (Fig. 4a and c).The highest output of papers was in 2017, with 35 papers published.The last decade produced 66% of all papers included in the corpus.Our corpus primarily consisted of laboratory studies, with studies consisting of both field and laboratory components being the second most common (Fig. 4b).The combined total of field and artificial studies comprised less than a quarter of all studies within our corpus.Land-use changes (n = 288) were the most commonly studied anthropogenic threat and occurred in every year since 1986 (Fig. 4c).Climate change first occurred within our corpus in 2010 and is the newest threat to be studied in relation to fish health within the 21st century, comprising 19 studies in total.The impact of water infrastructure has been studied for the longest period of time out of all our threat categories.Other threats ('Other', Fig. 4c) included impacts of artificial noise, artificial light and the impacts of bushfires, with only a single study related to each of these threats over the study period.Secondary stress responses were the most common health indicator reported, totalling 209 instances across all years and occurring in 37 individual years (Fig. 4a and c).Tertiary stress responses including whole-body morphology (122 occurrences across 33 years) and organosomatic indices (99 occurrences across 28 years) were the second and third most common health indicator (Fig. 4a and c).Other health metrics included assessments of skin mucous, otolith asymmetry and electrocardiograms, and were the least common, with six combined occurrences over the study period.
A total of 361 papers met the selection criteria for this review; however, a proportion of papers employed multiple health assessment methods, resulting in a total of 614 individual occurrences of fish health indicators.Figure 5 depicts these occurrences over a network of four stratum; Anthropogenic threats, Environmental alterations, Impact mechanisms and Health indicators, which when connected together by flows (coloured lines) form an impact pathway.Impact pathways can be followed either from left to right or right to left.Following from left to right uncovers the impact pathways a given anthropogenic threat follows to ultimately cause responses in fish health indicators.Following an impact pathway from right to left gives insight into the collections of threats that each health indicator is responsive to.Anthropogenic threats were dominated by land-use changes, largely due to a high proportion of papers relating to the impact of pollution from industrial, agricultural and urban inputs.As a result, increased chemical contaminants were the most common impact mechanism.In total, there were 81 unique impact pathways investigated within our corpus and despite land-use changes being the most common anthropogenic threat, water infrastructure resulted in the largest number of unique impact pathways (n = 29).Water infrastructure resulted in four different environmental impacts including thermal alteration, river impoundment, pollution and flow alteration.Land-use changes (n = 27) and climate change (n = 11) had the second and third highest number of unique impact pathways respectively.The threat of invasive species had the fewest unique impact pathways (n = 3), with the primary impact mechanisms being predation and changes in food availability, linked to only three health responses.These health responses can be seen in Fig. 6, which depicts the last two stratum of the main alluvial diagram for each anthropogenic threat to provide a clearer picture of the connections between impact mech-  to the impact mechanisms resulting from invasive species were behaviour, secondary stress responses and whole-body morphology (Fig. 6d).Impacts of artificial noise and light (categorized as other anthropogenic threats) were assessed using behavioural and primary stress response indicators (Fig. 6f).Increased chemical contaminants, largely driven by land-use changes, were the only impact mechanism linked to every health indicator recorded (Fig. 6c).Behaviour was the only health indicator to be measured in response to all our anthropogenic threat categories.The health indicators measured in response to the fewest anthropogenic threats were 'other' health indicators including assessments of skin mucous and otolith chemistry (assessed in response to climate change and land-use changes), as well as the assessment of lipids and fats (land-use changes and water infrastructure).

Discussion
This review compliments and extends on previous reviews (e.g.Brosset et al., 2021;Gomez Isaza et al., 2022;Stevenson and Woods, 2006) by focusing on the link between anthropogenic threats and individual fish health.By classifying papers based on a hierarchical framework of anthropogenic threats, environmental alterations, impact mechanisms and health indicators, we uncovered the impact pathways of six categories of anthropogenic threats to understand the mecha-nism by which these threats can impact fish health.We found that the impact pathways of these anthropogenic threats were complex.Multiple anthropogenic threats were related to the same environmental alterations and impact mechanisms.For example, studies related to the anthropogenic threats of water infrastructure and land-use changes both resulted in flow alteration (Fig. 6a and d).Overlapping of these anthropogenic threats leading to changes in hydrology has also been highlighted in previous studies and reviews (Poff et al., 2006;Poff and Zimmerman, 2010).Craig et al. (2017) refer to impact pathways as alteration profiles and define similar alteration profiles as having two or more shared environmental alterations.In most freshwater ecosystems, multiple anthropogenic threats are likely to be operating in tandem, and this can lead to combined alterations that can produce either additive, antagonistic or synergistic effects on fish health (Crain et al., 2008;Todgham and Stillman, 2013;Craig et al., 2017).A number of studies had anthropogenic threats that shared little overlap in environmental alterations and impact mechanisms.For example, studies in our corpus related to overexploitation followed a unique impact pathway, impacting fish health through fishing and subsequent handling.
In some instances, anthropogenic threats were directly linked to fish health metrics.Several articles directly linked thermal changes resulting from climate change to changes in fish health metrics such as secondary stress responses (Fig. 5  For panels (a) and (c), some papers within the corpus used multiple health indicators, meaning the number of occurrences shown in this figure is greater than the number of papers within the corpus.(Templeman et al., 2014;Kumar et al., 2020).It is likely that the impacts of climate change also have more indirect impacts in freshwater environments (Dudgeon, 2019); however, this finding highlights that anthropogenic threats can directly impact organisms within an environment instead of through complex pathways associated with environmental alterations and impact mechanisms (Watson et al., 2020).
Our corpus of articles was dominated by studies employing laboratory-based assessments of fish health.This is likely  The relative size of each category within each stratum is based on transformed (square root) occurrence counts to aid readability and is therefore not to scale.The plot is read either from left to right or right to left, with each stratum connected by 'flows', forming an 'impact pathway' across the four stratum.The proportion of each category within a stratum is provided numerically 'n'.related to the large proportion of studies assessing the impacts of pollution on fish health.Health components sensitive to pollution are typically related to primary and secondary stress responses, and most of these methods require laboratory analyses (i.e. the assessment of glucocorticosteroids within blood and tissue, the assessment of HSPs and the assessment of oxidative stress biomarkers (Sopinka et al., 2016).Contrary to our findings, in their review of the impacts of wildfires and associated runoff on aquatic fauna, Gomez Isaza et al. (2022) found that most studies they reviewed employed in situ (field-based) methods.Gomez Isaza et al. (2022) did not report on the types of health assessment methods that were attributed to in situ and laboratory studies; however, this difference in results suggests that the study type employed to assess the impacts of anthropogenic threats is highly dependent on the anthropogenic threat in question.In a review of management of wild fish populations in the Anthropocene, Cooke et al. (2022) highlight that field-based studies on a large number of individuals is likely the most valuable type of study for assessing responses in health to anthropogenic threats, and these types of studies are severely lacking within the literature.The comparatively low number of field-based studies in the corpus of this review further highlights this.
While our review focused solely on studies in the freshwater realm, many of the anthropogenic threats, their impact pathways and resulting fish health responses discussed in our corpus likely also operate in marine and transitional environments.In particular, Arthington et al. (2016) highlight overexploitation, climate change, the presence of invasive species and pollution as some of the most serious threats to marine fishes.These threats in marine and transitional environments can follow a multitude of direct and indirect impact pathways that mirror those in freshwater ecosystems   (Schull et al., 2023).For example, the impact of climate change resulting in increased thermal variation and extremes, causing primary and secondary stress responses, as well as the impact of land-use changes resulting in eutrophication, impacting a number of fish health components such as energetic reserves and immune response (Alfonso et al., 2021;Schull et al., 2023).
Finally, it is important to highlight that fish health responses can in some instances be sensitive to particular life stages and other intrinsic factors, meaning there are potential age-and size-specific considerations that should be made when investigating how fish respond to a stressor in question (Schull et al., 2023).Indeed, particular fish health indicators may be better suited to studying impacts of stressors at specific life stages.For example, RNA:DNA ratios have been recommended for use on juvenile or small-bodied fishes where body size may prohibit the use of other condition indices (Foley et al., 2016).Additionally, responses in fish health to stressors can be impacted by extrinsic abiotic features such as depth, salinity and other habitat characteristics (Schwartz et al., 2004;Zhao et al., 2013;Duarte et al., 2018).

Conclusion
Anthropogenic threats in freshwater environments pose a dynamic and complex problem; however, much of our knowledge relating to these impacts has focused on populationand community-level effects in freshwater ecosystems.Emerging fields such as conservation physiology seek to change this status quo and highlight the importance of individual organism health for overall population persistence and community stability.Our review identified key mechanistic links in the form of environmental alterations and impact mechanisms, through which anthropogenic threats impact fish health.Land-use changes were the most commonly studied anthropogenic threat.Anthropogenic threats impacted fish health through a diversity of impact pathways with water infrastructure consisting of the most unique combinations of impact pathways.We found that a large variety of fish health metrics are sensitive to these disturbances; however, secondary stress responses were the most commonly employed health indicator.We found that research related to freshwater fish health has been increasing over the past 50 years, with research dominated by laboratory studies.In light of our findings, future studies investigating anthropogenic impacts on individual fish health should seek to understand both the specific mechanisms and pathways through which anthropogenic threats may cause environmental alterations that impact fish health; as well as the fact that multiple anthropogenic threats can interact in unpredictable ways that can positively (additive or synergistic) or negatively (antagonistic) influence fish health.Linking threats to environmental alterations is crucial for ensuring effective management that addresses primary causes.We maintain that understanding drivers of individual fish health has practical implications for environmental man-agers, as individual fish health is the fundamental connection between anthropogenic threats and demographic changes at the population and community levels. ..........................................................................................................................................................

Figure 1 :
Figure1: A conceptual model showing the sequential steps through which stressors impact fish health, fitness and population dynamics.'Stressor' refers to anthropogenic threats and their resulting environmental alterations and impact mechanisms (described further on) that operate in freshwater environments.'Fish Health' refers to health components that make up the primary, secondary and tertiary stress responses.'Fitness' refers to parameters including but not limited to reproductive capacity, migratory capacity and survival.'Population dynamics' refers to factors such as recruitment, population resilience and biomass.The letters 'A', 'B' and 'C' highlight the links between these steps.

Figure 2 :
Figure2: A conceptual diagram depicting our framework of an impact pathway, linking anthropogenic threats to environmental alterations, impact mechanisms and fish health responses.The anthropogenic threat 'Water Infrastructure' is used as an example, which leads to flow alterations resulting in changed food availability, impacting fish condition as a tertiary stress response.

Figure 3 :
Figure 3: A global map showing the number of papers examining impacts of anthropogenic disturbances on fish health in freshwater systems published by each country.Grey polygons represent no papers published.

Figure 6 :
Figure 6: Six alluvial diagram subsets, one for each anthropogenic threat category from the main alluvial diagram in Fig. 5: (a) water infrastructure, (b) overexploitation, (c) land-use changes, (d) invasive species, (e) climate change and (f ) other.Each subset depicts the last two stratum, impact mechanisms and health indicators, of the main alluvial diagram in order to highlight the final stage of the impact pathway of each anthropogenic threat.Impact mechanism abbreviations are as follows (from top to bottom): CFA = changed food availability, CMP = competition, CPT = capture, CWQ = changed water quality, FS = fish screen, ICC = Increased chemical contaminants, II = Increased infections, MHC = morphological habitat characteristics, NL = noise/light, P = predation, R = radiation and TS = thermal stress.Health indicator abbreviations are as follows (from top to bottom): BEH = behaviour, EA = external abnormalities, GI = genetic indices, LF = lipids and fats, O = other, OI = organosomatic indices, PD = parasites and disease, PSR = primary stress response, SSR = secondary stress response and WBM = whole-body morphology.

Table 1 :
A summary of commonly measured primary, secondary and tertiary stress responses used as health indicators.Genetic indices and assessments of parasites and disease have been included as 'other'