Small instream infrastructure: Comparativemethods and evidence of environmental and ecological responses

1 Department of Biosciences, Swansea University, Swansea, UK 2 FreshwaterWorking Group, Society for Conservation Biology,Washington, D.C. 3 Institute forWater Research, Rhodes University, Makhanda, South Africa 4 Grupo de Investigación de Recursos Hídricos y Acuáticos, Universidad Regional Amazónica Ikiam, Tena, Ecuador 5 Centre de Ciència i Tecnologia Forestal de Catalunya, Lleida, Spain 6 River Restoration Centre, Cranfield University, Cranfield, UK 7 Cramer Fish Sciences–Genidaqs,West Sacramento, California 8 School of Aquatic and Fishery Sciences, University ofWashington, Seattle,Washington

5. Despite the abundance of road culverts greatly exceeding the number of small or large dams worldwide, they were evaluated in only 22% of studies that we reviewed.
Our findings underscore the need for studies to not only better understand local but also cumulative impacts of these smaller infrastructure, as these could be greater than those caused by large infrastructure depending on their location, density, and type, among other factors. Such studies are needed to inform infrastructure planning and watershed management.

K E Y W O R D S
dams, evaluation, freshwater ecosystems, rivers, roads, weirs

INTRODUCTION
Instream infrastructures such as dams, weirs, and culverts are widespread, and in many parts of the world continue to be constructed at unprecedented rates (Grill et al., 2019;Ibisch et al., 2016;Zarfl, Lumsdon, Berlekamp, Tydecks, & Tockner, 2015). Built for varied reasons, dams and weirs capture water and modify the magnitude and timing of its movement downstream, whereas culverts are constructed to facilitate the movement of water under roads and railways. Smaller infrastructures such as dams <15 m in height, weirs, and culverts are more prevalent and diverse in size than larger dams, yet are commonly neglected in environmental policy (e.g. Couto & Olden, 2018;Lange et al., 2019). It is estimated that there are 11 small dams for each large dam globally (Couto & Olden 2018), and the abundance of road culverts greatly exceeds the number of small dams (Fuller, Doyle, & Strayer, 2015;Januchowski-Hartley et al., 2013).
Despite recent attention given to smaller instream infrastructure, the diversity and pervasiveness of their environmental and ecological alterations across broad geographies remain poorly understood. A core limiting factor to enhance this knowledge is mixed regulation and laws for in-stream construction of infrastructure. There are notable

Types of infrastructure
Our study focused on dams or weirs <15 m high and culverts. Dams refer to infrastructure constructed along rivers by positioning a wall (spanning the channel cross section) intended to hold water back in a reservoir for different human purposes (e.g. water supply and hydroelectric power), and where flows are released downstream via different methods in a controlled manner (Richter & Thomas, 2007;Figure 1a). Weirs are like dams in that a structure is built across a waterway to transform conditions for different societal purposes (e.g. navigation and measuring water discharge; Figure 1b). But unlike dams, weirs often allow water to flow over the top of the structure. Culverts are structures whereby water from a river or other waterbody is diverted under a road, railway, or some other built structure (Truhlar et al., 2020; Figure 1c).

Literature review
We conducted a comprehensive search of ISI Web of Science (WoS) for articles published between 1972 and November 2017, with the following two sets of keywords: (a) (weir* OR low-head dam* OR run-of-river OR culvert* OR small dam) AND (impact* OR effect*) AND (enviro* OR eco*) and (b) (weir* OR low-head dam* OR run-of-river OR culvert* OR small dam) AND (water qual* OR water quan*). We used WoS because it references articles over a longer period compared to other databases such as Scopus (limited to articles since 1995) and returns more consistent results than Google Scholar (Nash & Graham, 2016). In using this  (Table S1). The 327 studies were then reviewed to determine (a) whether the study considered infrastructure (dams or weirs <15 m high or road culverts) specifically and (b) if the study was comparative in nature, evaluating environmental or ecological responses to infrastructure in reference to another system or condition. In total, 87 studies fit these criteria and were retained for detailed review.
Of the 87 studies we retained for detailed review, most were carried out in the United States (41), Australia (10), and the United Kingdom (8). From each of the 87 studies, we determined the types of infrastructure evaluated, how many, and the comparison method used.
In terms of comparison methods, we documented whether studies used spatial, temporal, or spatial and temporal comparative methods to evaluate environmental and ecological responses to infrastructure, and in two cases studies focused on modelling approaches that were spatially explicit and comparative. Specifically, we considered spatial comparisons as those that evaluated environmental and ecological variables in disturbed (infrastructure present) and undisturbed (no infrastructure present) waterways within the same or separate catchments, and temporal comparisons were considered as those that compared waterways before or after construction or destruction.
We also determined when a study used any combination of spatial or temporal comparative methods.
We also documented the frequencies of the different environ-

A note on infrastructure characteristics
A primary obstacle encountered through this review was a lack of both data on characteristics and reporting within studies; this influenced which studies were retained for further review. For example at least half of the 327 study publications that we initially reviewed were evaluations of dams or weirs, and of those roughly 80% did not report structure height. Some studies did include context or descriptive information that enabled us to make an assumption that dams or weirs were <15 m in height, but that was only the case for a small number of those included in our detailed reviews. Height, which we used as a characteristic for inclusion or exclusion from our analysis, is only one of several characteristics of infrastructure; it is also likely to be one of the more commonly reported. Explicit inclusion of these characteristics would allow representation of more studies in reviews such as ours and improve our understanding about how different typologies of infrastructure can alter and change freshwater ecosystems.
TA B L E 1 Summary of reviewed studies (N = 87) that considered single or multiple structures. All studies focused on a single type of infrastructure (e.g. dams), but could have conducted evaluations at more than one structure. We report the frequency (number of studies) that evaluated a single or multiple infrastructure type

Methods of evaluation
Most studies in our review evaluated responses at weirs (n = 37; 43%) or dams (n = 31; 36%). More than half (n = 43; 63%) of those evaluations at weirs and dams included more than a single structure (Table 1), meaning that a study focused on weirs could have evaluated multiple representatives of such infrastructure. We also found that nearly half (n = 36; 41%) of the studies evaluated between two and 10 structures.
Of the 87 studies that we reviewed, 10 different comparative methods were employed, 21% (n = 18) of which used multiple methods (Table 2). Spatial comparisons were used in 57% (n = 50) of studies, and within-catchment comparisons of disturbed versus undisturbed sites were most common (n = 36; 72% of spatial comparisons). Ten percent (n = 9) of studies used both spatial and temporal comparisons of environmental or ecological responses to infrastructure (Table 2).

Environmental and ecological responses to infrastructure
We found variable environmental and ecological responses (positive, negative, and neutral) both above and below different types of infrastructure, but there were several patterns that emerged (Figure 2a-f).
Notably, some studies evaluated more than one environmental or ecological response as well as responses above or below structures, resulting in 92 evaluations of environmental or ecological responses for dams in our study, 80 for weirs, and 43 for culverts (Figure 2; Tables S2-S4).
Overall, there were more evaluations of ecological (n = 111; 52%) compared to environmental (n = 104; 48%) responses to infrastructure in the studies that we reviewed (Figure 2a Table S3). Less than a quarter of evaluations above or below weirs found positive environmental or ecological responses relative to controls (Figure 2c-f). There were no positive or neutral environmental responses found above or below culverts (Figure 2e and 2f).

DISCUSSION
The widespread proliferation of infrastructure constructed along rivers calls for the need to develop a more robust understanding of associated impacts. Most studies in our review evaluated environmental or ecological responses at multiple dams or weirs and employed spatial comparative methods. We found that more than half of the evaluations that we reviewed reported negative environmental or ecological responses at dams, weirs, or culverts. Study evaluations also tended to focus on ecological responses to infrastructure, specifically on fish communities (just over a quarter). We discuss the implications of these findings below and outline recommendations for future studies, with F I G U R E 2 Summary of environmental and ecological variable responses (positive (blue), negative (orange), and neutral (grey)) above and below small instream infrastructure. Shown are frequency of variable responses (a) above and (b) below dams; (c) above and (d) below weirs, and (e) above and (f) below culverts. We report the frequency (number of evaluations) of environmental and ecological variables the goal to explore gaps in current knowledge and inform best practice for future evaluations.
Most evaluations in our review focused on dams or weirs, whereas culverts received less attention. This finding underscores that we have a limited understanding about the impact of culverts on freshwater ecosystems, and that there is a need for studies that include both large and small infrastructure (Grill et al., 2019). The scientific community should seek a more comprehensive, system-wide understanding about how infrastructure can influence and change freshwater ecosystems, especially because of their potentially large cumulative impact (Januchowski-Hartley et al., 2013). It was encouraging to find that studies in our review tended to evaluate environmental and ecological responses to multiple structures. However, our findings also suggest that there has been a tendency for studies to evaluate multiple larger infrastructure such as dams, but not necessarily multiple smaller infrastructure such as culverts. This could be the result of culverts not tending to occur in sequence along rivers, but that seems unlikely given their frequency along our waterways. It could also be that fewer stud- use single metrics such as taxonomic richness to assess ecological response, and in isolation these metrics can overlook changes to biodiversity (Mueller, Pander, & Geist, 2011). We suggest there is a need to move beyond examining such metrics in isolation to better understand ecological responses to infrastructure, and to measure taxonomic, functional, and phylogenetic properties for different biotic groups.
Such investigations should explicitly account for non-native species that are often found in higher diversity and abundance in reservoirs above dams (Johnson, Olden, & Vander Zanden, 2008). In relation to sampling, there could have also been differences in spatial and temporal scales at which evaluations were carried out, and those at which variables respond to infrastructure and associated changes in habitat and connectivity (Fullerton et al., 2010;Ganio, Torgersen, & Gresswell, 2005). While it is challenging to quantify environmental and ecological responses to infrastructure at all relevant spatial and temporal scales, there are methods (e.g. sensors, remote sensing, and machine learning) that can assist researchers with identifying the scales over which connectivity influences different ecosystem properties (Fullerton et al., 2010). Researchers should consider responses at nested spatial scales in relation to infrastructure along a river network and consider relevant temporal scales; this will depend on the response variable in question but needs to be given more specific consideration in future studies (Campbell, Lowe, & Fagan, 2007;Ward, Malard, & Tockner, 2002).
Roughly a quarter of studies that we reviewed focused on the ecological responses of fish communities to infrastructure. This is possibly why comprehensive reviews exist for fish (e.g. Fullerton et al., 2010)

CONCLUSIONS
Our review revealed a need to capture and record the characteristics of infrastructure more accurately. Doing so would facilitate our ability to scale up and extrapolate findings from individual studies to other areas (e.g. using results from a study on infrastructure over 5 m to estimate effects of infrastructure of the same height in other systems). In addition, our findings point to a need for studies that evaluate envi-ronmental and ecological responses to culverts, and this is particularly important because of common assumptions that these are less impactful on freshwater ecosystems despite occurring in higher densities than larger infrastructure. The relatively few studies that focused on culverts limit our ability to draw conclusions about patterns observed in studies. We see a need to move towards more comprehensive study designs, and we offered several directions that could foster a better understanding about how different variables are responding to infrastructure. Finally, most studies that we reviewed were conducted in three countries (the United States, Australia, and the United Kingdom), and we see that work is needed to understand how findings from wellstudied areas can help inform poorly studied areas. This is especially relevant as small infrastructure continues to expand around the globe.

AUTHORS' CONTRIBUTIONS
SRJ, SM, JC, VH, and SB conceived the study. SRJ, SM, JC, and VH designed the review procedure and gathered the literature. SRJ, SM, JC, VH, SB, and JDO reviewed the literature. SRJ, SM, and JCW analysed the data. SRJ wrote the article. All authors contributed to writing and editing and approved the manuscript for publication.

ACKNOWLEDGMENTS
The authors would like to thank the reviewers and editors for input that helped to improve the quality of this manuscript. We also thank the

CONFLICT OF INTEREST
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
The data used to produce the results reported in this study are available for download from Figshare: https://doi.org/10.6084/m9.figshare.