Review of Methods to Repair and Maintain Lithophilic Fish Spawning Habitat

: Rocky reefs provide important spawning and refuge habitats for lithophilic spawning ﬁshes. However, many reefs have been lost or severely degraded through anthropogenic e ﬀ ects like dredging, channelization, or sedimentation. Constructed reefs have been used to mitigate these e ﬀ ects in some systems, but these reefs are also subject to degradation which may warrant custodial maintenance. Monitoring and maintenance of natural or constructed spawning reefs are not common practices; therefore, few methodologies have been created to test the e ﬀ ectiveness of such tools. We conducted a literature review to assess available information on maintenance of rocky spawning habitats used by lithophilic ﬁshes. We identiﬁed 54 rocky spawning habitat maintenance projects, most of which aimed to improve ﬁsh spawning habitats through the addition of spawning substrate ( n = 33) or cleaning of substrate ( n = 23). In comparison to shallow riverine studies focused on salmonids, we found little information on deep-water reefs, marine reefs, or other ﬁsh species. We discuss the possible application of potential spawning habitat cleaning methods from other disciplines (e.g., treasure hunting; archeology) that may provide e ﬀ ective means of reef maintenance that can be used by restoration practitioners.


Introduction
Rocky reefs are historically important fish habitats that can increase fish abundance locally [1], serve as biodiversity hotspots [2,3], and provide spawning, nursery, and refuge habitats [4][5][6]. Habitat requirements vary by species and life stage (e.g., egg, larvae, juvenile, or adult); thus, different fishes and life history stages benefit from rocky reefs differently. Habitat requirements for egg and larval stages are constrained due to increased sensitivity to stochastic processes, smaller energy stores, and reduced mobility compared to adults [7]. For many species, clean rocky substrates in the influence recruitment potential in deeper waters [8]. Gravel beds serve the same purpose for salmonids and other riverine species in shallow water environments [9,10]. Despite their importance, many reefs and other rocky spawning areas have been lost or degraded due to habitat destruction from anthropogenic stressors [11,12]. Anthropogenic destruction of fish habitats is widespread and is implicated in >70% of North American fish extinctions and includes habitat loss or degradation or reduced access to spawning areas [13].
While marine coral reefs are the focus of many restoration initiatives, both marine and freshwater rocky reefs are not as well studied as coral reefs, despite having both high biodiversity and similar levels of degradation [14]. Causes of destruction or degradation of rocky spawning habitats include physical factors such as sedimentation [15][16][17], channelization [18,19], and resource extraction [20][21][22], while biological factors include periphyton accumulation (e.g., Cladophora or Didymosphenia) [23][24][25] and biofouling by invasive species (e.g., mussels [26,27]). Sedimentation is a natural process in freshwater and marine systems; however, anthropogenic stressors have increased the rate at which this process can occur [15,28]. Accumulation of fine sediments such as silt and sand can lead to egg mortality via infilling of interstitial spaces that help protect fish eggs and larvae from predators and displacement [29] while allowing for adequate water flow and oxygenation [17,30,31]. Biofouling occurs when the habitat is degraded through biological processes, such as colonization by Dreissena polymorpha and D. bugensis mussels (hereafter referred to as dreissenid mussels) or periphyton. Biofouling can also reduce the interstitial spaces required for egg incubation, deplete oxygen, and create waste build-up [27]. Many lithophilic spawners prefer substrate free of epiphytic growth and debris [29,32] and select spawning habitats with water velocities that maintain such habitats [33,34]. Historic destruction of reefs for resource extraction or development, coupled with altered ecosystem processes that increase rates of physical degradation, have led to the loss of viable fish spawning habitats and reduction of other ecosystem services [15,[35][36][37] (Figure 1). To mitigate impacts of degradation, construction of rocky reefs to improve spawning habitat increased during the 1980s [38] and spawning reefs have been constructed across the globe, with a majority located in North America and Europe [11,39]. Construction of reefs to attract fish has occurred in Japan for centuries, and construction of similar reefs began in the early 1800s in North America [25,40], but the emphasis on creating spawning-specific reef habitats has been more recent. To mitigate impacts of degradation, construction of rocky reefs to improve spawning habitat increased during the 1980s [38] and spawning reefs have been constructed across the globe, with a majority located in North America and Europe [11,39]. Construction of reefs to attract fish has occurred in Japan for centuries, and construction of similar reefs began in the early 1800s in North America [25,40], for revitalizing degraded constructed and natural rocky reefs. Considering the lack of existing research on spawning reefs specifically, the scope of this search included maintenance performed on all types of rocky spawning habitat due to similarities in the types of degradation and increased availability of the studied methodology. The specific objectives were to (i) compile and review a list of empirical studies which employ maintenance and physical monitoring of existing rocky spawning habitats, (ii) discuss causes of rocky spawning habitat degradation (physical and biological), and (iii) explore potential novel rocky reef maintenance techniques, including those from fields such as marine archeology and engineering.

Materials and Methods
A literature search was conducted for any available peer reviewed articles or grey literature (e.g., government reports, academic theses) published at any time using the search engines Google Scholar, Web of Knowledge, and Mendeley from August 2019 to April 2020. We searched the key terms "spawning reef", "rocky reef"," "spawning habitat maintenance", "habitat maintenance", "sediment removal", "reef cleaning", "gravel cleaning", and "reef monitoring", building the list as we reviewed the literature and identified new, more specific search terms (e.g., "algae treatment"). Coral reef studies were excluded from our search. Additional relevant material was explored by analyzing the citations within the found literature forward (literature which cited that manuscript) and backward (literature cited within that manuscript). The criteria used for including texts in the review were that they (i) focused on maintaining extant spawning habitats rather than the creation of new habitats, (ii) were based on empirical field studies, (iii) focused on lithophilic spawning fishes rather than other organisms such as invertebrates and aquatic plants, and (iv) included either natural or human-made habitats. Through the literature search, a list was compiled of habitat maintenance projects (i.e., a set of remedial actions taken in a given region) and relevant information including the targeted fish species, the type(s) of spawning habitat degradation, the remediation method(s) used to mitigate the degradation type(s), the years during which the studies were conducted or habitat maintenance occurred, the study area and study system type (lacustrine or riverine), and synopses of the maintenance method (e.g., area cleaned, amount of substrate added). The information was summarized by project rather than by each scientific study/manuscript because multiple studies were occasionally conducted on the same habitat maintenance project.

Results
There were 54 rocky spawning habitat maintenance projects found that met the criteria for inclusion in the literature review (Table 1). These efforts were summarized into categories based on the types of degradation and the maintenance/repair method employed, along with approximate estimates of scale and cost of the project in response (Table 2; Figures 2 and 3). Note that some projects had multiple types of degradation or maintenance methods, affecting the reporting of results (e.g., the total number of projects exhibiting each degradation type is greater than the total number of studies). Loss or removal of rocky spawning substrate (typically gravel) through high river flows, scouring, or other agents was the most common habitat degradation type (n = 30 projects), followed by sedimentation or infilling of substrate (n = 27) ( Figure 2). Other less common, and typically secondary, forms of degradation included channelization (n = 3), algae (n = 2), barriers to habitat (n = 2), and other biofouling (n = 1). Addition of substrate was the most common remedial action taken (n = 33) and was usually performed in response to loss of substrate (n = 27) ( Figure 3). Addition of substrate included gravel/cobble augmentation, installation of timber, and/or placement of boulders. Cleaning was the second most common maintenance method (n = 23) and was used in response to sedimentation (n = 23) or algae coverage or other biofouling (each n = 1). Methods employed to clean substrates included dredges, propeller wash, pressurized water, altered river flow, or mechanical methods (e.g., raking). Other forms of remediation included the addition of structures like gabions (n = 3), creation/placement of sediment traps (n = 4), removal of structures such as weirs (n = 2), and chemical treatments (n = 1). The majority of projects (n = 41) had no costs presented, and while (n = 13) did report costs, not all of them may have been complete.
The majority of rocky spawning habitat remediation projects focused on salmonids (n = 45), with few studies targeting walleye (n = 4), sturgeon Acipenseridae species (n = 3), or other fish species (n = 2), for a total of 25 fish species and one family (n = 26 total taxa). The most common individual species targeted were brown trout Salmo trutta (n = 14), chinook salmon (n = 13), and rainbow trout (n = 10; Figure 4). Rocky spawning habitat repair and maintenance efforts took place globally, but most occurred in North America (n = 38) and Europe (n = 13). Riverine environments (rivers, streams, creeks, etc.) were the most common habitats where projects took place (n = 52), with only two projects occurring in lacustrine environments such as lakes or reservoirs. The majority of projects took place in relatively shallow water, with 74% located in depths <1 m (n = 34 of 46 studies where depths were reported) ( Figure 5). Below, the maintenance methods employed in these studies are reviewed in the context of the physical, biological, and ecological scope of each problem and other potential remedial actions. In addition, an overview is provided of the physical monitoring techniques employed to monitor spawning habitat degradation and novel techniques that could be used in maintenance and remediation.       The majority of rocky spawning habitat remediation projects focused on salmonids (n = 45), with few studies targeting walleye (n = 4), sturgeon Acipenseridae species (n = 3), or other fish species (n = 2), for a total of 25 fish species and one family (n = 26 total taxa). The most common individual species targeted were brown trout Salmo trutta (n = 14), chinook salmon (n = 13), and rainbow trout (n = 10; Figure 4). Rocky spawning habitat repair and maintenance efforts took place globally, but most occurred in North America (n = 38) and Europe (n = 13). Riverine environments (rivers, streams, creeks, etc.) were the most common habitats where projects took place (n = 52), with only two projects occurring in lacustrine environments such as lakes or reservoirs. The majority of projects took place in relatively shallow water, with 74% located in depths <1 m (n = 34 of 46 studies where depths were reported) ( Figure 5). Below, the maintenance methods employed in these studies are reviewed in the context of the physical, biological, and ecological scope of each problem and other potential remedial actions. In addition, an overview is provided of the physical monitoring techniques employed to monitor spawning habitat degradation and novel techniques that could be used in maintenance and remediation.    Most common fish taxa (n = 10 of 26 total taxa) targeted by rocky spawning habitat maintenance projects. "Salmonids" represents studies where a target species of salmonid was not identified. Note that each project may have targeted more than one focus species.

Figure 5.
Water depths (m) and habitat types ("lacustrine" vs. "riverine") represented by the rocky habitat maintenance projects in our review with reported water depths (n = each water depth category represents the average or maximum water depth reported (pre-or post-maintenance)).

Discussion
This review highlighted several sources of degradation (e.g., loss of substrate, sedimentation, biofouling), each necessitating a different maintenance approach with options targeting the type of degradation to effectively revitalize rocky spawning habitat function. This literature review Figure 5. Water depths (m) and habitat types ("lacustrine" vs. "riverine") represented by the rocky habitat maintenance projects in our review with reported water depths (n = each water depth category represents the average or maximum water depth reported (pre-or post-maintenance)).

Discussion
This review highlighted several sources of degradation (e.g., loss of substrate, sedimentation, biofouling), each necessitating a different maintenance approach with options targeting the type of degradation to effectively revitalize rocky spawning habitat function. This literature review identified three primary types of degradation: loss of material, sedimentation, and biofouling by periphyton or invertebrates. Each type of degradation tended to be accompanied by a suite of tailored maintenance actions. In addition to methods reported from past studies, adapting methods from other disciplines (e.g., shellfish dredging, marine archaeology, treasure hunting and underwater recovery, marine engineering) may prove useful for maintaining reef function.

Loss of Material
Depletion of spawning gravel occurs when dams and diversion structures cut off or reduce sediment supply, preventing upstream gravel sources from replenishing downstream reaches [21]. Over time, optimal spawning substrates are transported away from spawning areas, leaving substrates that are too large for spawning fishes [112]. Due to this loss of gravel, remediation efforts often focus on the addition of spawning materials to supplement and replenish habitats. Gravel augmentation is commonly used to supplement depleted salmonid redds in western North America [44,79,85,113] but this technique has also been used to improve riverine walleye spawning habitats [74]. However, before applying gravel, there are important considerations such as substrate size, hydrology, and the frequency at which new gravel will need to be added or cleaned. While the removal of fine sediment could increase egg survivorship by improving the interstitial flow of oxygenated water [44], some studies have hypothesized that complete elimination of these small materials could damage embryos during sensitive developmental stages by inducing excessive flow velocity along with mechanical agitation [45].

Sedimentation
Sediment cover can negatively affect incubating eggs of lithophilic-spawning fishes [114][115][116][117]. Fine sediments within spawning substrates can reduce interstitial flow and dissolved oxygen, leading to decreased survival during early life stages [93,112,116,118]. Sediment compositions with over 12% fine substrates (<1 mm) are thought to reduce fry emergence by 50% in salmonids [112] and laboratory experiments found a 71% reduction in walleye egg hatch rates when covered by 2 mm of fine sediments [30]. Therefore, reducing sedimentation and fine sediments in rocky spawning habitats may be critical to successful recruitment out of the early life history stages in some locations.
Both active and passive sediment removal methods have been applied to reduce sediment accumulations on rocky spawning substrates. Active sediment removal methods, including cleaning of substrate through mechanical means, were more commonly employed in the habitat maintenance projects identified in our review (n = 23). In shallow water with shoreline access, heavy equipment such as bulldozers, walking excavators, or tractors have been used to rake or sift through substrate and dislodge sediment from impacted areas, allowing smaller particles to be washed downstream [86,93,103,105,111]. Variations of this method include use of an excavator equipped with a vibrating bucket that collects impacted gravel and sifts silt and sand away through a screened bottom [93]. A variety of gravel cleaning vehicles and devices that use high-powered water jets and/or suction mechanisms to dislodge or remove sediment have also been developed for use in shallow riverine environments. In Fraser River (British Columbia, Canada) spawning channels, a tractor-pulled, air-water jet system was used to flush fine sediment from substrate down to 50 cm below the riverbed [93]. Self-propelled vehicles such as the "Riffle Sifter" [91] and "Gravel Gertie" [94,95] drove along the river, directing vertical water jets into the streambed, removing fine sediments from down to 30 cm within the riverbed. A suction system then separated and removed the suspended fines, ejecting them onshore above the high water line. One pass could reduce the amount of fines by up to 78% [94,119]. A more modern invention for cleaning, the "Sand Wand", operates in a similar fashion as the vehicles, using water jets to dislodge fines and suction to remove suspended sediments, but instead of spraying sediment onshore, it is collected for later disposal. The unit is either float-or skid-mounted and has been used to treat several trout streams [106,108,109]. In a test of the equipment, 10 metric tons of sediment were removed from a study site at a rate of 1.2 m 3 of sediment per hour [108].
While the hydraulic cleaning methods used by the "Riffle Sifter" and "Gravel Gertie" have shown success in removing fine sediments, determining whether reductions in sediment resulted in an increase in salmonid egg to fry survival was never evaluated [119] and, in some cases, gravel cleaning did not guarantee spawning success [93]. Additionally, gravel cleaning via mechanical methods can dislodge eggs and others have detrimental side effects, including temporarily degrading water quality and distributing sediment to other areas downstream, which can disrupt insects and other aquatic invertebrates [91]. Non-target effects can be mitigated when applications are conducted outside of spawning periods and when stream flows are high enough to properly move sediment downstream [119]. Mechanical removal also requires obtaining site access for heavy machinery, which can be invasive (e.g., road construction and site clearing), and use of equipment along river corridors can reduce bank stability [120]. Therefore, the benefits of techniques to remove sediment can be weighed against the potential harm to spawning habitats to ensure that the maintenance efforts are practical and benefit the system with minimal collateral damage.
Dredging or flushing with high flows can be used to lift sediment from the target location, allowing the current to disperse the sediment downstream [121,122]. Dredging through suction methods can remove a large volume of sediment in a short amount of time. For example, hydraulic cutterhead dredges possess the ability to remove sediment at rates in excess of 1000 m 3 /h. However, due to the substantial disturbance that active dredging causes in the benthic area, it can create large turbidity plumes and resuspend large volumes of sediment into the water column [123]. This increased turbidity means dredging may not be practical in fragile habitats such as spawning gravel but may be applicable to habitats with larger substrates or where deep sediment layers persist. Passive methods of sediment removal include building sediment traps or collectors [96,99,102,109]. Sediment traps are large crevices excavated in front of rocky habitats, where the drop in depth causes the water velocity to decrease and excess sediment is allowed to settle [124][125][126]. While these passive methods are successful at collecting sediment from the water column, there have not been evaluations of their effects on spawning activity, egg survival, or early life history stage recruitment.

Periphyton
Periphyton is one of the most common biofouling agents, growing on the surface of rocky substrates if light penetration is adequate. Shallow freshwater reefs [127] and many nearshore areas experience some degree of algal growth (e.g., Cladophora) in the Great Lakes [128]. However, increased water clarity caused by the filtering action of dreissenid mussels has allowed for deeper growth in some areas, with Cladophora cover extending to 40 m deep on historically important lake trout spawning reefs in the Great Lakes [129]. Cladophora is commonly found on freshwater rocky spawning habitats, but little research has focused on other epilithic algae taxa despite their widespread nature [130]. During a study of two artificial reefs constructed in Lake Michigan, algal growth covered "nearly 100%" of the surface area of the shallower reef (9 m deep) within 1-2 years post construction, and while most of this growth was attributed to Cladophora, up to 131 species of periphyton were identified per year [25]. The deeper reef (11 m deep) experienced less algal growth, but algal coverage varied between 25-50% during later years [25]. Nuisance growth of algae has increased in many areas of the Great Lakes, the prevention of which is most effective at larger scales through nutrient management [131]. In systems with low nutrient levels in North America, Europe, and elsewhere, nuisance growth of the diatom Didymosphenia geminata can similarly blanket the benthos, causing disruption to food webs and covering valuable spawning habitats [23,132].
The presence of algae has been implicated as a factor leading to low reproductive output of lake trout [133] and lake sturgeon [134], but algae was not found to impact walleye spawning habitats in the Nipigon River, Ontario, Canada (not shown, 48.9755 • N, 88.2563 • W) [135] or Lake Erie [127]. However, lab experiments demonstrated that, compared to every other incubation substrate tested, the hatch rate of white sturgeon Acipenser transmontanus eggs on algae-covered rocks was the lowest (3-17% vs. >20-61% on most other substrates), which was attributed to fungal growth on the eggs [136]. Thick periphyton growth can form mats that cover entire spawning areas, preventing access to interstitial spaces and sometimes changing hydraulic conditions [23,132]. Sloughed mats of decomposing Cladophora or Didymosphenia diatoms can create anoxic conditions underneath the mat in as little as 3-6 h [137], which could decrease dissolved oxygen and suffocate developing embryos [132]. Periphyton growth on spawning habitats may also be detrimental to newly-hatched fry; Didymosphenia can entangle fish and limit food supply through the reduction of insect diversity [98].
Despite the widespread nature of periphyton growth on spawning substrates, periphyton removal as a habitat maintenance strategy has received little attention and determining the effectiveness of removal is difficult to characterize. One example is a shallow (4 m) reef constructed in the St. Lawrence River, USA that was abandoned by spawning fish three years post construction due to sedimentation, algal growth, and biofouling by dreissenid mussels [24]. To remove the fouling agents, Johnson et al. [24] used hand tools to clean a small area (9 m 2 ) and later used pressurized water to remove debris from a larger area (200 m 2 ). Despite cleaning efforts by researchers, the growth of algae and colonization by dreissenid mussels was determined to be too dense to attract spawning fishes [24]. Algae may be easier to physically remove in shallower river systems, where scarification with heavy equipment such as bulldozers is possible. In some human-made spawning channels for salmonids, the channel is dewatered in between spawning seasons and dredged and/or scoured with machinery to remove sediment and break apart debris such as algae mats [93,98]. The degree to which these techniques removed algae, as opposed to sediment, was not specifically measured. Other physical cleaning techniques employed in sediment removal (e.g., Sand Wand mentioned above) may also be effective at removing nuisance algae. Water flow management, such as increasing flows from dams, to scour away periphyton such as Didymosphenia may have limited application as this taxon thrives in relatively high water velocities (>1 m/s) [138] and anecdotal evidence found no loss in coverage after several storm surges [139].
Chemical treatments can be used to reduce nuisance levels of aquatic plants and filamentous algae, although their use necessitates a cost-benefit analysis and consideration for non-target effects. A combination of diquat and copper sulfide effectively removed Cladophora overgrowth in an artificial salmonid spawning channel off the Sacramento River in California, USA, although associated mortality rates of fish embryos and fry were not reported [63]. However, later chemical treatments using acrolein compounds in the canal resulted in massive fish kills and were discontinued [62]. While the chronic toxicity of diquat and/or copper sulfate is not well-studied [140], acute exposure to these chemicals can harm desirable native plants [141], reduce survival of tadpoles [142], and reduce survival of rainbow trout Oncorhynchus mykiss in the embryo/alevin stage [140]. Nuisance Didymosphenia blooms are a more recent phenomenon and chemical treatment protocols are under development [138,143]. A field test of one chemical compound effectively reduced coverage of the diatom but resulted in 16-20% mortality of rainbow trout caged in the treatment area [143]. Alternative methods of algae treatment involve placing barley straw or barley straw extract (antifungal) over the affected area [144] or placing glass beads covered in titanium dioxide, which breaks down chloroplasts in a reaction with sunlight [145]; however, both will likely only have localized effects. Prevention of biofouling by stopping the introduction and spread of nuisance organisms may be necessary to protect isolated spawning habitats. Public campaigns to sterilize boats and fishing equipment may be effective for preventing the spread of invasive species, including Didymosphenia, to new areas [146,147]. Continued experimentation with methods to control and remove periphyton is needed, but incorporating mitigation for biofouling during reef restoration planning, design, and construction may be the best practice to minimize potential for degradation impacts.

Invertebrates
Infestation of rocky surfaces by invasive mussels is a common problem in North America and Europe [148] that may limit the reproductive success of lithophilic spawning fishes [27]. Dreissenid mussels, for example, were first discovered in North America in the 1980s and quickly spread throughout the Great Lakes, with live mussel densities of 100,000/m 2 typical in some nearshore areas [148]. Most rocky areas and reefs in the Great Lakes are now colonized by live mussels and dreissenid shell fragments [27,55,129,[148][149][150] and their colonization has led to lake-wide changes in nutrient dynamics and energy flows (e.g., the "nearshore shunt") [151]. Furthermore, invasive Pacific oysters Crassostrea gigas in Europe have also spread quickly since the 1980s and colonized rocky habitats [152,153].
Mussel colonization of new substrates such as constructed spawning reefs can occur quickly; reefs built during 1986-1989 in Lake Erie had 75-80% coverage by 1991, just three years after zebra mussels were discovered in the area [150], and reefs in southern Lake Michigan were covered 4 years post construction [27]. Spawning habitats colonized with mussels have reduced interstitial space, which may deter or prevent fish from spawning on fouled reefs [149,154]. Large accumulations of shells from dead mussels may also decrease interstitial space, but the shells are similar in size to gravel substrates and may be used as spawning substrates by some fishes [55]. Like overgrowth of some benthic algae, mussel beds and their pseudofeces may reduce oxygen availability for developing fish embryos. Dense mussel beds can heavily influence oxygen dynamics in the surrounding substrate: trapped feces/pseudofeces can form anoxic zones, and hypoxia forms atop and within mussel beds when combined with Cladophora growth [137]. Often, one invasive species can facilitate the spread of more nuisance species [155,156], and dreissenid mussels have increased the biomass of Cladophora algae through water filtration (thus increasing light levels) and the addition of nutrients [151]. Preliminary results from Lafrancois and Bootsma [157] show that the removal of mussels may reduce the occurrence of algae on reefs as well.
Despite the widespread invasion of mussels globally, there are limited reports about the direct effects of dreissenid mussels on spawning success. Fitzsimons et al. [158] did not observe differences in walleye spawning activity, egg deposition/survival, or interstitial water quality on Lake Erie reefs that had been fouled by mussels at extremely high densities (>300,000 m 2 ), but their results were not comparable to pre-invasion spawning data. In a comparison of unfouled cobble and dreissenid encrusted spawning habitats in southern Lake Michigan, lake trout egg deposition and survival were much higher on the clean substrate, with many damaged eggs found on the mussel beds [27]. Binder et al. [34] observed lake trout spawning activity on northern Lake Huron sites with clean gravel/cobble substrates but also on lower quality substrates, such as areas fouled with mussels and algae, smaller substrates, and large boulders. Reduction of substrate quality related to mussel coverage is considered an understudied aspect of lake trout management, and researchers have suggested measuring egg survival on manipulated substrates (cleaned vs. fouled) [159]. Building spawning reefs has been recommended as a strategy for lake trout recovery in the Great Lakes, with the caveat that the construction accounts for potential mussel fouling [160].
In this review of rocky spawning habitat maintenance methods, studies detailing the removal of mussels from spawning habitats were not found, but this is likely because efficient large-scale treatments have only recently been tested or implemented. Treating larger infestations of mussels is complicated due to greater possible effects on non-target organisms. Water drawdowns of impounded lakes are effective at decreasing nearshore concentrations of mussels [161] but can also harm nearshore fauna such as native unionids or herpetofauna [162] unless they are transplanted. Chemical treatments may be one of the most effective ways to remove larger dreissenid infestations from small lakes or targeted areas of large lakes. A number of molluscicides are effective against dreissenid mussels, including niclosamide, potassium chloride (usually in potash), and copper compounds (e.g., cupric ions in EarthTec QZ), but all have been shown to have non-target effects on native unionid mussels except for Zequanox, which specifically targets dreissenids, using dead soil bacteria Pseudomonas fluorescens to degrade their digestive lining [163].
The widespread nature of mussel invasions makes management of the problem difficult, especially in large lake ecosystems. However, mussel prevention and removal techniques are being field-tested, and some targeted removal strategies may be employed to maintain rocky spawning habitats. Aquaculture facilities can also prevent the spread of veligers (i.e., larval dreissenid mussels) by treating water containing fish eggs [164,165] and juvenile fish [166] during transport to other water bodies. In newly colonized areas, manual removal of small mussel beds by divers is labor intensive but may be possible to prevent further spread, such as in Lake George, New York, USA (not shown, 43.4262 • N, 73.7123 • W) [167]. Volunteer divers and biologists are currently testing this technique on spawning shoals in northern Lake Michigan to measure the effects of mussel removal on water quality and algae growth, removing~1 million mussels during 2016 [157]. Small patches of mussels can also be removed using water jetting at 3000 psi (e.g., Figure 6A), although this may be limited to applications such as hull and dock cleaning unless detached mussels can be collected and removed from the area [168,169]. Additionally, adaptation of a tillage disc cultivator offers the opportunity to crush and dislodge the mussels without destroying the reef (e.g., Figure 6B). Depriving mussels of oxygen by covering mussel beds with tarps or by pumping carbon dioxide over mussel beds can also target smaller mussel accumulations, although these have yet to be employed in many field applications [170,171].
can also be removed using water jetting at 3000 psi (e.g., Figure 6A), although this may be limited to applications such as hull and dock cleaning unless detached mussels can be collected and removed from the area [168,169]. Additionally, adaptation of a tillage disc cultivator offers the opportunity to crush and dislodge the mussels without destroying the reef (e.g., Figure 6B). Depriving mussels of oxygen by covering mussel beds with tarps or by pumping carbon dioxide over mussel beds can also target smaller mussel accumulations, although these have yet to be employed in many field applications [170,171]. Figure 6. Potential reef cleaning mechanisms, including (A) pressure washing by divers using tools for cleaning large vessels, (B) removal of mussels or periphyton using a large vessel and a pasture or disc harrow, (C) excavation of sediment with a prop wash diverting mailbox system for large vessels (depths up to 15 m) and small vessels (water depths <5 m), (D) sediment removal via pumps or engines such as those used by personal watercraft.

Novel Remediation Techniques
Our results suggest a need to explore multiple disciplines for remediation methods, especially for deep-water and/or lacustrine environments, which were underrepresented in our review. For example, marine archeologists commonly use hydro suction technology such as airlifts or suction dredges to remove sediment by a vacuum-like process, allowing the substrate below to remain intact [172][173][174]. While effective on a small scale, this methodology has a small footprint and is labor intensive. Thus, hydro suction methods may require additional modification to be an efficient Figure 6. Potential reef cleaning mechanisms, including (A) pressure washing by divers using tools for cleaning large vessels, (B) removal of mussels or periphyton using a large vessel and a pasture or disc harrow, (C) excavation of sediment with a prop wash diverting mailbox system for large vessels (depths up to 15 m) and small vessels (water depths <5 m), (D) sediment removal via pumps or engines such as those used by personal watercraft.

Novel Remediation Techniques
Our results suggest a need to explore multiple disciplines for remediation methods, especially for deep-water and/or lacustrine environments, which were underrepresented in our review. For example, marine archeologists commonly use hydro suction technology such as airlifts or suction dredges to remove sediment by a vacuum-like process, allowing the substrate below to remain intact [172][173][174]. While effective on a small scale, this methodology has a small footprint and is labor intensive. Thus, hydro suction methods may require additional modification to be an efficient method for maintaining large spawning reefs. Propeller wash from marine vessels is another method to flush fine sediments that is less precise but has a larger target area than hydro suction methods [175][176][177]. Using a "mailbox" device, turbulent water is directed downward by the propulsion of the vessel's propeller and suspends fine sediments from the benthic substrate; however, this requires a vessel large enough to produce adequate propulsion to sufficiently reach the target depth ( Figure 6C). Additional novel methodologies that could potentially be adapted for sediment removal include long-tail motors, bow thrusters, and even personal watercrafts (PWC) with modified propulsion units ( Figure 6D). Since these methods are portable and deployable from a range of vessels, they could provide effective methods for maintaining rocky habitats in both lentic and lotic systems.
Fine sediment removal methods were recently tested in lentic environments using two benthic sled devices that utilized either pressurized water or propulsion-driven water to clear sediments and other fouling from the interstitial spaces of two rocky spawning reefs in Saginaw Bay, Lake Huron, USA [58]. These devices were small in scale (1.35 m × 1.0 m; <90 kg) and were able to be towed by a smaller research vessel (6.4 m long). Results showed that both cleaning devices were able to successfully increase the relative hardness of the bottom substrate in some sampling areas, suggesting that the cleaning devices have the potential to remove sediments from rocky spawning reefs [58]. Furthermore, removal of sediments and increases in relative hardness over cleaned areas of the reef resulted in increased egg deposition by walleye and lake whitefish in these areas. Though these results are just one example of novel reef cleaning methodology being tested, the development of cleaning devices and subsequent results of cleaning (e.g., increased relative hardness of cleaned areas and increased egg deposition) provide support and future guidance for the implementation of this method.
Prevention of sediment accumulation through mechanical means may be possible in some lotic environments. While techniques such as adding riparian vegetation have been used extensively to control runoff and sediment loads [178,179], recently developed "active sediment collectors" have been installed in several Michigan, USA salmon streams and have had some efficacy in preventing the deposition of bedload sediments [109,180], although additional field testing and quantification of sediment collection would be beneficial to assess their efficacy in spawning habitat maintenance. The sediment collector is a ramped box placed perpendicular to the flow across the river bottom, which allows for the settling and collection of sediments that are periodically suctioned onshore for disposal. Laboratory testing of a small (0.6 × 1.1 m) active sediment collector found bedload reductions of 93-99% for grain sizes >0.5 mm and reductions of 54-75% for smaller-grain materials under a variety of flow conditions [181]. Field testing of a larger (9.14 m wide) collector in Fountain Creek, Colorado, USA (not shown, 40.5163 • N, 87.8094 • W) was undertaken to determine whether the system could reduce the need for dredging operations in the system, and the collector obtained maximum sediment collections of 76.5 m 3 /h under high flow regimes, removing an estimated~670,000 m 3 of sediment per year with minimal maintenance [182]. More conceptual methods of sedimentation prevention include the development of a plan for a "self-cleaning substrate" or a system of tubing placed under spawning substrate that uses the hydraulic head of the river to push water under and through the substrate, potentially dislodging accumulated particles [183]. Although never field tested (D. Geiling, Fisheries and Oceans Canada, personal communication), this technology could be considered for testing in the creation of new habitat projects or could be retooled for use in lentic habitats by using pump action instead of river head to create flow through the system. Development of novel sediment removal devices for fisheries should include input from biologists and rigorous field testing, as scale models may yield different results. This was the case with the movable baffle developed for the Tehama-Colusa Canal, CA, USA, which was designed to lower into the water to create high flows and clean sediments [92] but was largely ineffective at scale, creating unsuitable patches in the spawning habitat, blocking fish access, and frequently breaking down [62].
Large-scale chemical treatments to remove and destroy invasive invertebrate species have been tested with success; however, until more treatments are conducted to analyze their effects on native species, other methods may be necessary for large-scale removal. One such example that we propose is a modified pasture harrow ( Figure 6B), used to break apart and distribute oysters on reefs, that can be dragged across the site by boat to dislodge and crush nuisance mussels or break apart mats of epiphytic algae or diatoms without displacing reef structure. Both boat-towed disc and tooth harrows have been tested for use in oyster cultivation [184], and specialized toothed raking apparatus or "mud cleaning machines" have been developed to scatter accumulated sediment and clustered shells [185]. Conventional dredging methods and hydraulic excavators have also been used to exhume oyster reefs buried by sediment [186,187]. Published literature about the use of harrows and other oyster reef cleaning mechanisms is scant but contact with local managers about adapting techniques used may prove useful to maintaining lithophilic fish spawning habitat as well.

Physical Monitoring Techniques
Physical monitoring of spawning habitats is often necessary to determine whether maintenance is warranted, and, if the habitat has been constructed or remediated, monitoring is needed in order to determine whether the project is meeting management objectives. Identifying possible problems in a system before construction begins allows managers to plan and manage for degradation before it makes the site unusable. Biological monitoring of remediated habitats is common, but for spawning reefs constructed in the Great Lakes only 24% of projects reviewed included some form of physical monitoring post construction [54]. Through these monitoring efforts, researchers observed changes in reef placement [188], physical damage from logs and ice jams [189], or sedimentation/biofouling [27,55,[189][190][191]. Furthermore, similar degradations have been observed in marine systems [192,193], highlighting the need for incorporating metrics of salmonid habitat improvement projects reporting any type of project structure into monitoring.
The type of physical monitoring needed depends on the management goals and the type of environment; different protocols will be needed for physical monitoring in shallow vs. deep-water areas or lentic vs. lotic environments. Designing a physical habitat monitoring program should consider which physical variables are important to the study species and which physical processes in the study system may act on the habitat to degrade its effectiveness over time. Key elements to consider monitoring are the physical structure of the habitat, hydrology, and water quality. Furthermore, research on potential elements to include in physical monitoring programs for deep-water reefs include focusing on measuring the following metrics: water depth, tide, current magnitude and direction, wave height, reef position, and condition. Monitoring is suggested to be conducted 6 months after initial deployment and following up every 12 months from then on [194].
Methods for monitoring the spawning habitat's physical structure (e.g., substrate, placement, structural integrity) will depend on the project's goals, budget, and expertise in collection and post-processing of data. Loss or infilling of substrate is one of the most important elements to monitor, and traditional methods for measuring substrate size (e.g., Ponar or Eckman grabs) may not be effective in rocky, deep-water environments or may represent too small of an area relative to the size of the habitat. In situ measurements of substrate size are complicated in deep-water environments, and divers may be needed to map habitat structure, count organisms, or assess conditions, either by placing quadrats [27] or by following transects [195]. In shallow or deep water, stable reference points (such as rods) can be placed to determine whether rock pile height has changed [196]. Sediment collection devices can be used to measure deposition rates over time [127]. In lieu of in situ measurements of habitat structure, underwater video (UTV) cameras towed from a boat, hauled along dive transects, or carried by remotely operated vehicles are commonly used reef monitoring tools [55,190,[197][198][199]. Obtaining quantifiable data from UTV can range from relatively simple techniques (e.g., calculating substrate size and area fouled from still frames) [55] to using computer vision/deep-learning algorithms to delineate substrates and map habitats [198], vegetation [200], or invertebrates [201]. Side-scan sonar and multibeam bathymetry are other advanced technologies that allow researchers to map substrates and benthic structures that are commonly used to assess fishes' habitats [55,202,203]. These advanced technologies are an effective suite of tools for assessing reef condition. Standardizing methods and protocols for monitoring rocky reefs will further improve their application and future efforts to compare reef maintenance methods across systems.
Biological monitoring of remediated habitats is common, but for spawning reefs constructed in the Great Lakes, only 24% of projects reviewed included some form of physical monitoring post construction [54]. Another example shows that only 6.7% of salmonid habitat improvement projects report any type of monitoring [204], as discussed in the previous review. In addition, a review assessing whether or not these spawning habitat projects are effective found that only 4% of all projects evaluated were considered successful. A majority of these studies did not qualify for success because of the 3-year post-monitoring requirement to be considered effective [61]. This lack of monitoring creates a situation in which success and effectiveness cannot be evaluated.

Conclusions
This review of rocky spawning habitat maintenance shows that maintenance of rocky spawning habitats and reefs is not a common practice, with few standardized techniques developed, including a lack of information regarding cost information and evaluating success. Increased monitoring and reporting of costs and effectiveness will improve the ability of practitioners to make informed decisions on proceeding with maintenance projects. This effort identified that most habitat maintenance projects focus primarily on riverine habitats and salmonids, highlighting the need for more research conducted on deep-water species like lake trout, cisco Coregonus artedi, and sturgeon species. We also found a dearth of publications detailing rocky reef maintenance in marine habitats, despite the high rates of sedimentation observed on marine reefs such as those off the coast of Italy and Turkey [205], but this may be due to our bias in selection of English language publications and search terms. Although maintenance through substrate addition has the potential to be successful for habitats suffering from loss of substrate, each habitat is unique and, while gravel augmentation is the most common remediation technique, it might not be uniformly applicable for all target species or life stages. For habitats that experience an influx of too much material, i.e., sedimentation, adding gravel is not a viable solution; therefore, when considering potential methods of maintenance, the solution should incorporate the source and mechanism of degradation. Adapting and experimenting with methods from other disciplines (e.g., shellfish dredging, marine archeology, treasure hunting and underwater recovery, marine navigation, and dredging) may prove useful for repair and add value to these reef restoration programs. Furthermore, cleaning systems and anticipated maintenance needs can be incorporated into artificial reef construction designs and budgeting to facilitate routine maintenance and increase the long-term success and function of spawning reefs. Finally, our review shows that post-management monitoring and shared results through scientific outlets could be beneficial to all those involved in rocky reef restoration and maintenance, including managers, researchers, and practitioners.