Phytoremediation using aquatic plants

12.

Freshwaters are affected by a diverse range of pollutants which increases the demand for 5 effective remediation. Aquatic phytoremediation is a nature-based solution that has the potential to 6 provide efficient, spatially adaptable and multi-targeted treatment of polluted waters using the 7 ability of macrophytes to take-up, sequester and degrade pollutants. This chapter considerers the 8 primary phytoremediation mechanisms that macrophytes employ to remove inorganic, organic and 9 biological waterborne pollutants before highlighting some of the common macrophyte accumulators 10 that have been studied. Three common macrophyte planting systems (i) constructed wetlands 11 (CWs), (ii) wild macrophyte planting/harvesting and (iii) floating treatment wetlands (FTWs),are 12 considered to understand how macrophytes are deployed for targeted aquatic phytoremediation. 13 Important practical considerations for implementing aquatic phytoremediation include the 14 use of invasive species, the optimal harvesting time and frequency for pollutant removal with 15 macrophyte biomass, and the full extent of the role that microbial biofilms play in phytoremediation. 16 In this chapter, these issues are unpacked and recommendations for future programmes of research 17 and development are made. Finally, the opportunities to generate 'added value' from expanding 18 aquatic phytoremediation in terms of the provision of ecosystem services and the potential for 19 resource recovery are outlined. 20 12.2Water contamination and water security 21 22 Surface waters are vital for supporting people and ecosystems; however, freshwater 23 availability is under increasing pressure due to a growing human population requiring access to safe 24 water (Heathwaite, 2010). Global freshwater resources comprise 2.5% of the total global water 25 budget, although only 0.0072% (93,120km 3 ) of the total global waters are available for drinking, 26 energy, food production and the industry sector (Lawford et al., 2013;Zimmerman et al., 27 2008). Tilman et al.(2011) predicts that crop production will need to increase by 100-110% by 2050 28 to feed the growing population, leading to a global freshwater deficit of approximately 2,400km 3 per 29 year (Rockström et al., 2014). 30 Many surface waters are currently of sub-optimal standards due to a range of stressors 31 impacting freshwaters such as point source and diffuse pollution, land-use change and climate 32 change, which further compounds the challenge of providing water security (Ormerod et al., 2010;33 Berger et al., 2017).One of the major pressures on water quality in the United Kingdom is nutrient 34 enrichment from diffuse pollution (Ulénet al., 2007), whereas elsewhere in countries such as 35 China,additional issues of heavy metal pollution are also prominent (Cheng, 2003). Interactions 36 between different stressors in space and time can also lead to additive effects (Heathwaite, 2010), 37 for example, increased land-use change towards intensive agriculture and a potential increase in 38 storm frequency may increase the delivery of nitrogen (N) phosphorus (P) and fine sedimentto 39 receiving water (Dunn et al., 2012). 40 Table 12.1 summarises the surface water pollutants that are of concern and where 41 remediation solutions are being developed. Water pollutants can be broadly categorised as either: 42 organic, e.g. hydrocarbons, pesticides and algal toxins, orinorganic,e.g. metals or syntheticand 43 manure-based fertilisers containing excess amounts of N and P,or biological,e.g. pathogens and algal 44 toxins.The mobilisation and effects of different pollutants have been discussed extensively 45 elsewhere (Heisler et al., 2008;Ohe et al., 2004;Liess & Carsten Von Der Ohe, 2005;Edwards, 2015;46 Lintelmann et al., 2003). However, different pollutants may have multiple sources, for example, N 47 and P can be released from agriculture, aquaculture and urban waste water streams. 48 Macrophytesuptake and sequester N primarily in the form of nitrate (NO3 -) and ammonium (NH4 + ), 172 while P is taken up as phosphate (PO3 4− ).Studies vary in their focus on total amounts (i.e. including 173 particulate) versus the dissolved fraction of macronutrients,which makes comparing optimal 174 macrophyte accumulator specieschallenging (Table 12.3). Macrophytes that have the greatest 175 biomass production and/or fastest growth rates are some of the most effective 176 nutrientphytoremediators (Keenen and Kirkwood, 2015), for example, Eichhornia crassipes, Lemna 177 sp. and Typha latifolia have growth rates of 60-110 t/ha/yr, 6-26 t/ha/yr and 8-61 t/ha/yr, 178 respectively (Gumbricht, 1993). 179 Emergent species have received considerable attention in nutrient phytoremediationand are 180 often deployed in CWs, with Canna spp. and Cyperus spp. showingsome of the highest removal 181 efficiencies for ammonium (NH4 + ) of between 74-100% (Table 12.3). Typha latifolia, Lolium 182 multiflorum and Polygonum hydropiperoides showed high TP removal efficiency of 81-90% (Table  183 12.3).For floating macrophytesEichhornia crassipes, Lemna gibba and Pistia stratiotes show good 184 potential for nutrient removal:E. crassipescan remove up to 92% NO3and 81% NH3whilst L. 185 gibbacan remove 100% NO3and 82%NH3 - (Table 12.3). The same two species were also effective at 186 removing total phosphorus (TP) (Table 12.3). Submerged plants have received less attention for their 187 nutrient phytoremediation capacity (Table 12.3). This may reflect the difficulty of cultivating and 188 harvesting submerged macrophytes, and thepotentially lower biomass generated compared to 189 emergent plants (Du et al., 2017). Ceratophyllum demersum and Myriophyllum aquaticumare 190 potential candidates for the targeting of total nitrogen (TN) and TP with removal rates >41% (Table  191 12.3). Potamogeton crispuswas deployed as part of a hybrid FTW experiment and was found to have 192 enhanced effects over the FTW comprised of only emergent plants; however, the individual removal 193 contribution from P. crispus was not quantified (Guo et al., 2014). Most submerged species are 194 rooted in sediment and may also remove nutrients from the water column through foliar absorption 195 (Eichert and Fernández, 2011). Hence they offer the dual ability to remove nutrients from water and 196 sediment,allowing the simultaneous remediation of sediments that have a pollutant legacy and 197 which may continue to release nutrients to the water column via internal loading even after external 198 loads have been reduced. However, the disturbance caused during harvesting can re-199 suspendsediment-bound elements, and alter the macrophyte-equilibrium state to a potentially 200 undesirable phytoplankton-dominated state (Kuiper et al., 2017). 201 The phytoremediation potential of a macrophyte is influenced by biotic factors such as 202 competition, predation anddevelopmental stage (Quilliam et al., 2015), and abiotic factors such as 203 temperature, pH, light availability, seasonality and nutrient loading (Ansari et al, 2014). For example, 204 Ayyasamy et al.(2009)found that the removal efficiency of byE. crassipesincreased between 205 concentrations of 100mg/l to 300mg/l of NO3 -, but decreased at higher concentrations of 400 and 206 500 mg/l of NO3 -. Similarly, a mesocosm-based study of the effect of different temperature regimes 207 on N and P removal byNasturtium officinale and Oenanthe javanica found that maximum net 208 accumulation of TN and TP occurred at an air temperature of 22°C but deteriorated thereafter (Hu et 209 al., 2010). Given the wide range of factors that may influence the ability of macrophytes to remove 210 contaminants, understanding the performance of some of the key macrophyte accumulators under 211 different environmental conditions is prudent in order to optimise species selection. 212

214
Macrophytes canalso remove micronutrients(henceforth referred to as metals (Rai, 2009))from 215 water and sediments, and hyperaccumulators are most appropriate for the phytoremediation of 216 metals (Ali et al., 2013). The search for hyperaccumulator species has been one of the primary foci 217 within the field given the widespread prevalence of past and current metal industrial effluents and 218 the ecological risks they carry (van der Ent et al., 2013); however,metal bioavailability can be reduced 219 by sedimentation and adsorption to clay particles (Kumar et al., 2008). Studies based on mesocosm-220 scale CW experiments have been carried out on synthetic solutions with elevated metal 221 concentrations in domestic and industrial wastewaters to assess the potential of macrophytes of 222 different growth forms to act as hyperaccumulators (Fu & Wang, 2011;Kamal et al., 2004;Rai, 2009;223 Rezania et al., 2016) (Table 12.4). Many species also have the capacity to take up multiple types of 224 metals meaning that some species could be more beneficialin phytoremediation (Table 12.

4). 225
Macrophytes that have often been cited as hyperaccumulators with high biomass potential are 226 free-floating plants, such as members of the Lemnaceae (e.g. Lemna minor), Pista stratiotes, 227 Eichhornia crassipes and those from the generaSalvinia (Table 12.4). For example, L. gibbahas been 228 reported to concentrate between 14,000mg/kg dry weight of Cd, whilst E. crassipes can concentrate 229 10,000mg/kg Zn (Low et al., 1994;. Furthermore, Typha latifolia and 230 Cetatophyllum demersum L. have also shown good potential (Osmolovskaya & Kurilenko, 2005;231 Sunita et al., 2015). The main limitation for macrophyte metal uptake is the toxicity of the target 232 metal pollutant at higher concentrations (Landesman et al., 2011). However, detoxification 233 mechanisms also allow species to avoid the negative effects of these metals (Deng et al., 2004); for 234 example, more than 50% of the Ca, Cd, Co, Fe, Mg, Mn, and Zn recovered in the roots of Pistia 235 stratioteswere actually attached to the external surfacesindicating theability of the plant to exclude 236 metals and thus maintain tolerable levels internally (Lu et al., 2011). Newete & Byrne(2016) also 237 state that the extent of the root system affects the ability of macrophytes to remove metal 238 pollutants, with fibrous root systems being superior due to their large surface area.Physio-chemical 239 factors are also important for uptake and accumulation of metals with temperature, light, pH and 240 salinity all having been shown to influence remediation performance (Rai, 2009 whilstMyriophyllum aquaticum can remove 76% and 82% respectively (Gao et al., 2000). Elodea 253 canadensisalso has the ability to remove 48% to 89% of p,p'-DDT (Gao et al., 2000;Garrison et al., 254 2000). Lemna gibba, L. minuta and Potamogeton crispus have been demonstrated to be very 255 efficient at removing phenols from water (Barber et al., 1995;Hafez et al., 1998). However, P. crispus 256 is less efficient at removing twoPAHs, phenanthrene (removal 18-34%) and pyrene (removal 14-24%) 257 (Meng et al., 2015). 258 There is great potential for phytoremediation of a wide variety of PPCPs such as anti-259 inflammatory, hormonal replacement and anticonvulsant products . CWs(section 260 12.7.1) planted with Phragmites australis demonstrated very efficient removal of the hormones 261 Estrone, 17 beta-estradiol and 17 alpha-ethinylestradiol from water (Table 12.5). In CWsthe water 262 column/plant sediment matrix adepthof c.7.5cm provided more efficient PPCP removal than 263 deeperdepths of 30cm . This highlights the importance of oxygen for the removal 264 of waterborne hormone pollutants with vertical mixing from the surrounding atmosphere increasing 265 the aeration of plant roots and . Plants such as Typha latifolia with more 266 extensive roots and rhizomes system may be favourable for deployment due to their capacity to 267 oxygenate water (Makvana and Sharma, 2013). 268 Scirpus validus displays mixed ability to remove anti-inflammatory pharmaceuticals with 269 very efficient removal of naproxen, compared to very poor removal of diclofenac (Zhang et al., 2012;270 Zhang et al., 2013a). Typha angustifoliaremoved 27-91 % of anti-inflammatory drugs in a study by 271 .  found that there is large variability in planted rural CWs in 272 terms of their removal efficiency of PPCPs with 11-100% removal of anti-inflammatories, 37-99% for 273 β-blockers and 18 -95% for diuretics. Understanding this variability and identifying macrophytes for 274 the removal of PPCPs through laboratory studies and at the field-scale is important given the need 275 for lowcost removal solutions, especially in developing countries. There has been little focus on the 276 use of novel macrophyte planting systems(e.g.FTWs) for the removal of organic chemicals, and 277 future work on these systems would build flexibility into the deployment of different aquatic 278 phytoremediation schemes for tackling the problem of PPCP pollution. Importantly, the distribution 279 and storage of organic chemicals within plants, especially for PPCPs, requires further study in 280 orderto avoid the problem of transferring pollutant from one place to another(sections 12.8 and 281 12.9). 282

284
Most studies on the removal of microbial pollutants and their indicators of the presence 285 (e.g. E.coli,faecal coliforms and faecal streptococci)are focused on macrophytes within CWs, 286 therefore the following examples will mainly refer to this planting type (see section 287 12.7.1).Furthermore most studies show that CW planting systems removemicrobial pollutants from 288 watervia a combination of chemical, biological and physical mechanisms.A study of 12 CWs found 289 that over a year vegetated CWs removed between 95-97% of faecal coliforms and 93-98% of faecal 290 streptococci (Karathanasis et al., 2003). Similarly, in an experimental CW system, Makvana & 291 Sharma(2013) Phragmites australisshowed no significant difference in the removal rates (>98 %) between the two 295 treatments; furthermore,in general, unplanted mesocosms reached their maximum removal rate 296 before the planted mesocosms(with the exception of theC.alternifoliusmesocosm) suggesting that 297 plants provide little additional benefit for removing biological pollutants over and above the effect of 298 standing water conditions (Kipasika et al., 2016). Similarly, a review comparing Lemna sp. treatment 299 ponds against unplanted treatment ponds showed that the latter had greater removal rates of 300 E.colifacilitated by the greater exposure of the water to UV light and the subsequent 301 photodegradation and microbial die-off (Ansa et al., 2015). However, Decamp & Warren (2000)have 302 shown that gravel beds planted with Phragmites australisremove E.coli more quickly compared to 303 unplanted soil beds, possibly as a result of the impact of antagonistic root exudates from P. 304 The variability of the results obtained between planted and unplanted experiments suggests 306 that for each treatment system different mechanisms of microbial pollutant removal become 307 dominant. Within unplanted facultative systems or lagoons it is likely that oxygenation and 308 phytodegradation from UV light are the dominant methods of removal (Ansa et al., 2015). 309 Conversely, biological and chemical process may become more important within planted systems, 310 for example, Pistia stratiotesfacilitates presence of protozoa by providing structural habitat, which 311 can increase predation on Salmonella (Awuah, 2006). Conversely, predation from protozoa seemed 312 to have a negligibleeffect in systems planted with Spirodela polyrhiza (greater 313 duckweed),highlighting that removal mechanismsare probably related to below-ground 314 morphological attributes, with more extensive roots/rhizomes providing superior habitat for grazers 315 (Awuah and Gyasi, 2014).Increased root zone surface area also facilitates greater microbial biofilm 316 growth which is thought to be a key removal structure for bacterial adsorption and predator 317 microbial proliferation (Decamp and Warren, 2000). Therefore,smaller grasses such Festuca 318 arundinacea may have limited potential for microbial pollutant removal compared to large emergent 319 such asTypha latifolia (Decamp and Warren, 2000). Futureresearch investigating the ability of 320 different macrophytes to remove microbial pollutants from water, especially outside of CWsystems, 321 is clearly merited. Direct deployment of macrophytes for pathogen removal would be highly 322 beneficial in developing countries where low-cost options for remediation could provide accessible 323 water treatment. 324 Of the few experimental studies investigating potential for macrophyte removal ofmicrobial 325 pollutants outside of CWs, Saeed et al.(2016)  induced by experimental 'shock phases' where hydraulic loading was increased between 5 to 14-fold 328 to simulate low frequency and high magnitude discharge events, the removal of E. coliwas reduced 329 significantly to levels varying between 6-45%. The effect of hydraulic retention time is also important 330 for pathogen survival and die-off (Reinoso et al. 2008)and may have implications for the use of 331 phytoremediation (with FTWs) in lakes and rivers given the difference in hydraulic retention times. 332

334
There has been considerablework focusing on the ability of individual plant species to remove 335 single pollutants from water(e.g. Zhou & Wang 2010), with the design of CWs also focusing on 336 monocultures of macrophytes (Kadlec, 2009). Conversely, there has been a lack of studies that 337 explicitly explore the ability ofmixedplant assemblages to simultaneously take-upand degrade 338 multiple pollutants (Koelbener et al., 2008). A plant community-based approach provides the 339 opportunity to enhancethe removal of both single pollutants, but also target multiple 340 contaminants.Studies that have looked specifically at phytoremediation using plant communities 341 have shownencouraging results (Fraser et al., 2004;Liang et al., 2011;Türker et 342 al., 2016). For example, an experiment testing the removal of N and P from four different emergent 343 macrophytes in parallel (Carex lacustris, Scirpus validus, Phalaris arundinacea and Typha latifolia) 344 found that microcosms planted with all four macrophytes in equal proportion, either matched or 345 outperformed microcosms planted with a single species (Picardet al.2005). Earlier studies also 346 suggest thatplant polycultures have a greater removal potential for heavy metals and can reduce 347 biochemical oxygen demand (BOD) (Karpiscaket al., 1996;Scholes et al., 1999). However, Türker et al. 348 (2016) reported that boron removal from mine effluent was more effective in native emergent 349 monocultures compared to polycultures, although the opposite was true for NO2removal. These 350 results suggest that there are probably optimal plant combinations for particular pollutants and 351 further experiments designed to identify these combinations would help to optimise the efficiency 352 of phytoremediation. 353 Toassembleappropriate plant combinations there are several importantfactors to consider 354 including the functional diversity of the community. It has been reported that simply increasing 355 species diversity in a plant assemblage can increasenutrient removal, although polycultures 356 containing more thanthree species showed no further benefit (Ge et al. 2015;Geng et al. 2017). A 357 common theme among these studies is the importance of species identity in explaining variation in 358 nutrient removal, where specific combinationscan more effectively remove pollutants. Therefore, 359 assembling appropriate plant communities based around the complementaryphytoremediation 360 potential of individual species, and the interaction of those plants withothers in the assemblage is 361 potentiallymore important than simply increasingspecies richness per se. However, the effect of 362 competition between plants is important to recognise as this may impact the community 363 composition, and therefore the ability to remove the targeted pollutants from water (Zhang et al. 364 2007). In a mesocosm experiment, containing the submerged macrophytes Stuckenia pectinata 365 (Sago pondweed), Potamogeton natans (broad-leaved pondweed), Potamogeton crispus (curled 366 pondweed) and Zannichellia palustris (horned pondweed), it was found that S.pectinata reduced the 367 biomass of the other species (Engelhardt & Ritchie, 2001). Reducing the biomass of certain species 368 willnot necessarily compromise overall removal efficiency as uptake and sequestration potential will 369 vary with species. However, this highlights the need to understand interspecific interactions in order 370 to enhance removal efficiency, especially when considering targeting water bodies in a non-371 equilibrium state where conditions favour the dominance of one particular species (Engelhardt & 372 Ritchie, 2002). 373 A field studyemploying plant communities revealed some of the benefits of combining multiple 374 macrophytes (Wang et al., 2009;Zhao et al., 2011). Nine macrophytes species (five floating, one 375 submerged and three emergent) deployed on FTWs and planted on river banks outside Jiaxing City, 376 China, demonstrated removal rates of TN and TP at 16%-37% and 26%-43% respectively (Zhao et al., 377 2011). Although the removal rates were relatively low,it was also highlighted that the plant 378 community-based approach allows for species within the community to compensate for deficits in 379 uptake of other species . For example, the average P content of floating 380 macrophytes was ca. 5.9g/m 2 , whereas, emergent species including Canna indica and 381 Pontederiacordata with higher biomass accumulation, stored P at a level ofca. 7.3g/m 2. .Similarly,a 382 phytoextraction study with emergent species (Carex flava, Centaurea angustifolia and Salix caprea) 383 allowed the impact of facilitation across increasing concentration gradients to be seen (Koelbeneret 384 al., 2008). Here, the willowS. caprea attenuated the toxic effect of Zn on therelative growth rate of 385 C. flava by lowering the availability of Zn, thus mitigating the negative effect of Zn on the 386 sedge (Koelbeneret al., 2008). This highlights that competitive effects may not always be negative 387 and may produce positive effects through 'over yielding'. The consequences of competitive 388 interactions between candidate macrophytes evidently deserve particularattention within the field 389 of plant community-based phytoremediation. 390 As well as the potential enhanced removal of pollutants from plant communities with 391 macrophytes of different life forms (Koelbener et al., 2008) there may also be the potential for 392 generating ecosystem services from polycultures.A 2-yearstudy by Wang et al. (2009)explored the 393 potential restoration of Lake Taihuand Lake Machou byusing a mosaic of macrophytes in 394 successional stages highlighting the potential for spatial and temporal diversity in macrophyte 395 deployment, and the provision of ecosystem services. Floating and emergent macrophytes were first 396 introduced to reduce light availability for algal growth,facilitating theintroduction of submerged 397 species leading to removal rates of TN and TP of 60% and 72% (Wang et al., 2009). The provision of 398 ecosystem services due to the different plant life forms was highlighted as an advantage by Wang et 399 al.(2009)as increasedpatches of vegetation provided refuge for zooplankton that subsequently 400 grazed phytoplankton.The added value of diverse plant communities is a factor that requires 401 quantificationto espouse the benefits of aquatic phytoremediation over and above water treatment. 402 Plant community-based approaches provide the opportunity to build temporally more 403 consistent treatment into phytoremediation by exploiting the differing phenology of plant species; 404 polyculture systems canthus offerthe most consistent water treatment option with least 405 susceptibility to seasonal variation (Karathanasis et al., 2003). However, the temporal dynamics of 406 plant communities within the context of phytoremediation are under-researched, and there is a 407 needto explore the assembly of plants, e.g. in terms ofdiffering phenologies,to extend the growing 408 season, especially in temperate regions where water treatment potential declines after senescence. The most effective phytoremediators have fast growth rates and high biomass 414 accumulation; however, outside of their native range macrophyte species with these traits are often 415 considered to be invasive, and given their potential for rapid colonisation they can quickly 416 outcompete native macrophytes (Chambers et al., 2008). Species that are invasive in the UK, such as 417 Azolla filiculoides and Hydrocotyle ranunculoides, can clog waterways and have serious ecological 418 impacts on native flora and fauna (Schultz and Dibble, 2012). In the UK, the combined cost of 419 controlling invasive plants, together with their economic impact, is estimated to be £1.7 billion per 420 annum(The Great Britain Non-native Species Secretariat, 2015). Therefore, there is a 421 significantjuxtaposition between using species of invasive plants in phytoremediation, and 422 management strategies to control invasive species (Rodríguez et al., 2012). Given that in many cases 423 the complete eradication of invasive aquatic macrophytes such as Eichhornia crassipes is unlikely, it 424 may be more appropriate to exploit these macrophytes as part of an integrated management 425 strategy that controls the spread of these species whilst at the same timeeffectively removing 426 nutrients and metals, capturing suspended sediment, and harvesting the biomass for economic gain 427 (Patel, 2012;Yan et al., 2017). A similar parallel can be drawn with non-native and invasive zebra 428 mussels (Dreissena polymorpha) which are often considered detrimental (Matsuzaki et al., 2009), but 429 have also widely been reported to stabilise the clear-water state of shallow lakes through filtering 430 phytoplankton and removing harmful cyanobacteria (Gulati et al., 2008). 431 Water bodies where invasive species are already present may be targeted for active 432 harvesting allowing periodical regrowth for continued phytoremediation (Xu et al., 2014). However, 433 there are important factors to consider including the containment of macrophytes to avoid 434 transferto other water bodies (e.g. via contaminated harvesting equipment or through downstream 435 spread of fragments), including the most appropriate harvesting technique, and the sustainability of 436 exploiting such an ecological engineering systems (Rodríguez et al., 2012;Yan et al., 2017). The site-437 specific context will likely determine the appropriateness of active harvest of invasive aquatic plants 438 (Yan et al., 2017). In terms of introducing macrophytes into a freshwater system for 439 phytoremediation, it is inappropriate, and indeed possibly illegal, to deploy invasive species given 440 the potential for ecosystem damage and long terms effects. In these circumstances non-invasive or 441  (Kivaisi, 2001;Nivalaet al., 2007;Tanner, 1996;Vymazal, 2009;456 Vymazal, 2011).CWs also show potential for treating wastewater containing emerging contaminants 457 of concern including pharmaceuticals and other endocrine disrupters (Vymazal, 2009). 458 CWs can be categorised as free water surface flow wetlands (FWSF) or sub-surface flow (SSF) 459 wetlands (Dhir, 2013) (Figure 12.3). FWSF wetlands containemergent, floating and submerged 460 macrophytes growing in shallow ponds or lagoon watersover sandy or organic soils, which allows the 461 influent contaminated water to slowly flow through the emergent macrophyte stems for maximum 462 pollutant uptake and UV degradation (Kadlec, 2009). SSF wetlands are the most common type of CW 463 and comprise emergent macrophytes growing over a substrate of stone or gravel matrix enabling 464 water to comein direct contact with plant roots, rhizomes and biofilms,which promoteaerobic 465 conditions (Vymazal, 2011). Several processes including physical filtering of the water, biological 466 processing of water by plants and microbial biofilms, and chemical changes due to redox state can 467 assist in pollutant removal in SSF systems (Faulwetter et al., 2009). The average SSF CW system is 468 100 times smaller than the FWSF CW system (Kadlec, 2009), therefore, FWSF are more common in 469 North America and Australia where a larger surface is available, whilst SSF wetlands are more 470 common in Europe where land availability is more limited (Vymazal, 2011). SSF wetlands are 471 frequently used to ameliorate the concentration of biologically derived organic material as indicated 472 by the lowering of biochemical oxygen demand (BOD) and chemical oxygen demand (COD) from 473 waste waters (Vymazal & Kröpfelová, 2009). 474 CWs are the most advanced form of macrophyte deployment within the umbrella of aquatic 475 phytoremediation (Kennen and Kirkwood, 2015). However, these systems can require high 476 investment costs and they are restricted primarily to pollutant point sources where there is 477 wastewater treatment such as tertiary sewage treatment and wastewater polishing before water 478 enters a natural waterway (Patiño Gómez and Lara-Borrero, 2012). This restricts the application of 479 CWs for the treatment of water containing pollutants from diffuse sources. Although CWs have the 480 potential to be utilised for treatment of a wide range of contaminants, their most widespread 481 application has been for sewage wastewater-related contaminants, including BOD, COD, N and P, 482 and often they are set up with crop monoculture to maximise plant uptake (Kadlec & Wallace, 2009;483 Sundaravadivel & Vigneswaran, 2001;Vymazal, 2009). 484 CWs vary in level of design and engineering required for their development; FWSF wetlands 485 are generally low tech gravity-fed systems, whereas, SSF require more construction and 486 management to import the stone/gravel matrixes, and also may include bundsto separate different 487 treatments then requiring the use of electric pumps (Kadlec and Wallace, 2009). In both types of 488 CWs there are high investmentsin construction and operational costs. CW can also become clogged 489 with sediment, which impacts the functioning of the system and imposes additional costs for 490 excavation and removal of contaminated sediments, and the subsequent reinstatement of 491 macrophytes (Machado et al., 2016). According to design guidance for the treatment of urban waste 492 water and sewage, SSF CWs may require an area of around 5m 2 to 10 m 2 of CW per person 493 equivalent for adequate water purification (Tilley et al., 2014). Therefore, given the potentially large 494 area required, CW-based phytoremediation may be unable to compete for limited land availability 495 with other more profitable land uses. Furthermore, in countries where vector-borne diseases, such 496 as malaria or dengue, are a public health issue the creation of open shallow wetland environments 497 may be undesirable as it has the potential to provide ideal conditions for the propagation of 498 mosquitoes and other disease vectors (Mwendera et al., 2017). 499 From both industry-based observations and from the available literature, the primary purpose 500 of CWs is water treatment and wastewater polishing. This however, ignores their potential to offer 501 ecosystem services such as sequestering and harvesting nutrients for reuse, provisioning for 502 biodiversity, pollination and carbon sequestration, and thus underplays the overall value of CWs. 503 There is great potential to develop different post-remediation 'streams' which have been relatively 504 unexplored, and which emphasise support for different ecosystem services (see section 12.10.2). 505 Aquatic phytoremediation is a promising technology for the treatment and remediation of polluted 506 water with the operational point-source based CW systems in place, but given the limitations of 507 these systems, including the lack of application for diffuse pollutants, investment costs and lack of 508 ecosystem focus there is an opportunity to further develop context-specific, sustainable 509 phytoremediation that provides ecosystem services within wider environmental systems. 510 12.7.2 Wild macrophyte harvesting 511 512 Most aquatic phytoremediation planting systems involve the deliberate deployment (FTW) or 513 engineering of planted systems (CWs). Harvesting of existing wild macrophytes from water bodies 514 such as shallow lakes can also be a phytoremediation strategy,and relies upon the opportunistic and 515 timely removal of macrophyte biomass in order to manage waterborne pollutants suchas N and 516 P (Huser et al., 2016). A study of an urban shallow lake, showed that harvesting an annual amount of 517 3,600 kg dry weight of Elodea canadensis led to 16.4 kg P being removed from the system, equating 518 to around 53% of the TP load removed (Bartodziej et al., 2017). Although the estimated cost of 519 removal was $670 per kg of TP, which was more expensive than chemical flocculating treatment,this 520 wasstill considerably less expensive than many catchment best management practices (Bartodziej et 521 al., 2017). Macrophyte harvesting is often carried out in lakes and waterways ostensibly to relieve 522 navigation, drainage, aesthetic or recreational problems, rather than for phytoremediation 523 purposes, but is notable that nutrient export may be a collateral benefit of such harvesting. Other 524 case studies have shown that macrophyte harvesting for nutrient removal does not reduce nutrient 525 loading quite as favourably (Carpenter and Adams, 1977;Morency andBelnick, 1987), withPeterson 526 et al. (1974) estimating that plant harvesting only removed 1.4% of TP loading. 527 The variation between these case studies is possibly a result of the levels of nutrient loading, 528 with waters that receive extremely high inputs of nutrients leading to a poor offset by removal from 529 plant harvesting (Bartodziej et al., 2017). Another source of variability for nutrient removal is the 530 coverage of macrophytes across the particular water body; the reported optimal coverage of 531 macrophytes ranges from 5% to 40% (Portielje and Van der Molen, 1999;Dai et al., 2012;Xu et al., 532 2014). For environmental managers considering macrophyte harvesting as a mechanism for in-water 533 nutrient management, it is crucial that a scoping study is carried out to determine the base balance 534 of nutrient input/output and plant removal capacity, and to identify the need for upstream best 535 practices as part of an integrated management strategy. 536 The harvestingmethod itselfis also an important element of harvesting wild macrophytes, e.g. 537 removal by hand, or mechanically via specialised boats equipped with cutting or raking apparatus 538 (Quilliam et al., 2015). Hand removal is labour and time intensive, although it allows targeted 539 macrophyte removal and minimises disturbance (Quilliam et al., 2015). Conversely, mechanical 540 removal allowsmore rapid and extensive removal but is non-selective and can lead tohigh levels of 541 turbidity due to the re-suspension of sediments. This can impact invertebrates and fish by removing 542 structural habitat and may ultimately drive the system from a desirable clear water macrophyte-543 dominated state to a potentially unfavourable phytoplankton-dominated state (Dawson et al., 1991;544 Sayer et al., 2010;Habib and AR, 2016). 545 In some circumstances it may be necessary to establish macrophytes in waterbodies by direct 546 planting through seeding or transplanting propagules (e.g. tubers/root crowns) if there are 547 noexisting macrophytes, or if a particular species is required to target certain pollutants (Smart et 548 al., 1998;Hilt et al., 2006). In addition to plant establishment there is also scope to 549 enhancemacrophytegrowth and biomass by engineering interventions such as the assembly of 550 polytunnels over vegetation, or enclosures to reduce grazing losses. 551 12.7.3 Floating treatment wetlands 552 553 Within aquatic phytoremediation one such novel ecological engineering solution that has been 554 developed is the FTW. The premise of this system is that highly productive emergent macrophytes 555 such as Typha latifolia are planted within a growth medium, which is supported by a buoyant frame 556 allowing the roots of the emergent macrophytes to be submerged in the water, thus enabling 557 rhizofiltration, phytoextraction and phytodegradation to take place hydroponically (Nichols et al., 558 2016;Kiiskila et al., 2017) (Figure 12.4). Root uptake associated with FTWs is primarily applicable to 559 water-soluble contaminants within the water column only, although sediment-bound pollutants 560 canbe physically filteredfrom the water column by plant roots (Tanner and Headley, 2011b). FTWs 561 have recently gained increased attention and may also be referred to in the literature as artificial 562 floating islands, integrated ecological floating beds, floating plant bed system and hydroponic root 563 mats (Yeh et al., 2015). 564 FTWs can accommodate fluctuations in water levels, andthe stability of materials used to 565 construct the buoyant frame may include items such as polyvinyl chloride (PVC) pipes, foam sheets, 566 bottles and bamboo (Ladislas et al. 2013;Wang et al. 2015;Pavlineri et al. , 2017). However, it would 567 be useful within the literature if qualitative information and design challenges were alsoreported to 568 provide an idea of performance and usability of FTWs in practice, and although there are no 569 reported incidences of FTWs capsizing or other failures during pilot tests, this may simply reflect 570 publication bias. 571 Netting material or foam is generally used to support the growth medium in which the 572 macrophytes are grown (Yeh et al., 2015). Material previously used as substrate includes peat, soil, 573 cotton and coir fibre (Pavlineri et al., 2017). Furthermore, FTWs comprising foam with gaps to 574 support pots have also been designed (Lynch et al., 2015). Growth media physically supports the 575 planted macrophytes and provide nutrition, but the substrate can also enhance pollutant removal 576 through the stimulation of microbial activity (Tanner & Headley, 2011a). Macrophytes may be 577 established by transplanting of seedlings, cuttings or whole plants (Yang et al., 2008;Ning et al., 578 2014). An advantage of using FTWs rather thandirect plantingof macrophytes is the ease in which 579 the biomass can be harvested from the frame, instead of having to remove plants from the 580 sediment. The quick and simple method of harvesting afforded by growing plants in FTW facilitates 581 recovering pollutants from plant biomass (Bartodziej et al., 2017). There is potential for quick re-582 planting of the FTW for continued remediation and biomass removal Ge et al., 583 2016). 584 FTWs have been studied principally for their capacity to remove nutrients, but there have also 585 been attempts to assess heavy metal, pathogen and phytoplankton removal (Borne, 2014;Yeh et al., 586 2015;Jones et al., 2017;Kiiskila et al., 2017). FTWs have been deployed at a variety of different 587 scales including microcosms, mesocosms, and as pilot trials within lagoons (Headley and Tanner, 588 2008;Ladislas et al., 2013;Chang et al., 2014;McAndrew et al., 2016;Nichols et al., 2016;Kiiskila et 589 al., 2017). Here the experimental polluted water used has included storm water, lake water, river 590 water, sewage effluents, domestic wastewaters, refinery wastewater, acid mine drainage, and 591 livestock effluents Li et al., 2012;Borne, 2014;Wang and Sample, 2014a;Abed, 592 Almuktar However,they may not be representative of the real remediation performance given that polluted 598 waters contain a multitude of chemicals and microbes which may influence remediation (Javadi et 599 al., 2005). Therefore, further studies would benefit from testing the remediation of water sourced 600 from the environment. 601 Only a small handful of field-scale experiments have been carried out that assess the usefulness 602 of FTWs in successfully remediating pollutant-impacted waters McAndrew et al., 603 2016;Nichols et al., 2016;Olguín et al., 2017). Of the available studies that assessFTW performance 604 within water bodies, including streams, urban and rural ponds, results focus on plant tissue element 605 accumulation rather than the arguably more pertinent issue of water quality improvement (Zhu et 606 al., 2012;Olguín et al., 2017;McAndrew et al., 2016;Nichols et al., 2016). Although plant tissue 607 sequestration is extremely important for assessing the bioaccumulation potential of macrophyte 608 species it does not explicitly demonstrate water quality improvement; this can only be proven 609 through monitoring water chemistry. Scaling up mesocosm scale experiments to assess actual field-610 scale water quality improvement is challenging given the ideal of a control site with comparable 611 water chemistry and abiotic and biotic conditions, or high-temporal resolution baseline water quality 612 data forthe experimental water body, both of which may be unavailable. Where there is a clear 613 opportunity for upstream and downstream water quality sampling near the experimental FTWs, 614 such as a stream, water quality changes are more likely to be attributed to the FTW intervention 615 between these points (Olguín et al., 2017). Similarly, more field studies longer than 2 years, ideally 616 up to 5 to 10 years, would lead to a better understanding of the longer-term performance of FTWs 617 and, crucially, reveal the actual remediation time (Yang et al., 2006).Furthermore, the influence of 618 inter-annual hydrological variability on FTW performance in terms of precipitation and evaporation 619 could also be evaluated. Despite the paucity of scientific studies at the field scale, commercial 620 companies now commonlyoffer FTWs as a water treatment solution, and as part of the aesthetic 621 enhancement of urban rivers. The phytoremediation research community must aim to keep pace 622 with the private sector to corroborate industry-advocatedbenefits of FTWs and avoid any potential 623 reputational damage to aquatic phytoremediation where expectations of these systems from 624 stakeholders are not met (Keenen and Kirkwood, 2015). 625 The remediation performance of FTWs is highly variable with reported minimum and maximum 626 removal efficiencies for TN values being 0.71 mg/l (4 %) and 51 mg/l (91 %) and 0.06 mg/l (1 %) and 627 18.85 mg/l (90 %) for TP (Figure 12.5). This high variability may be due to differences in FTW design, 628 macrophyte species employed, and the chemical composition of the experimental water. A further 629 example of variation in removal efficiency comes from Lynch et al. (2015)who compared two 630 commercial FTWs (Beemat and BioHaven®) planted with the rush Juncus effusus that had been 631 designed to treat storm water. It was found that Beemat FTW outperformed BioHaven® in both TN 632 and TP removal (Lynch et al. 2015). The difference in removal may have been due to the difference 633 in substrate (coir matting vs. sphagnum peat) or the physical design of FTW (Lynch et al. 2015).The 634 growth medium is indeed an important source of variability within FTW design. Rice straw used as 635 growth medium was found to enhance removal of TN, NH4 + and NO3compared to plastic filling (Cao 636 and Zhang, 2014). Similarly, the FTW with straw filling had a greater total densityof nitrifying and 637 denitrifying bacteria which suggests thatthisorganic material was providing both a habitat and a 638 source of C for the growth of microorganisms, which were able contribute to pollutant metabolism 639 (Cao and Zhang, 2014).Commercial FTWs are still an expensive management option, and there is 640 currently a demand for more low-cost growth media that both provides a suitable substrate for 641 macrophytes and enhances pollutant removal; such examples includebiochar, activated carbons, 642 coffee waste and green compost (Tran et al., 2015). To date there has been no research 643 incorporating these materials into FTWs to assess the potential for enhanced remediation and the 644 potential value post-remediation. 645 HybridFTW planting systems are being developed in an attempted to enhance pollutant removal 646 and ecosystem restoration (Guo et al., 2014;Li et al., 2010;Lu et al. 2015). Such systems integrate a 647 new layer beneath the floating platform containing submerged macrophytes such as Potamogeton 648 crispus, and/or bivalves such as freshwater clams (Corbicula fluminea) (Guo et al., 2014;Li et al., 649 2010) (Figure 12.6). Photovoltaic solar panels have also been attached to the frames of FTW to 650 power a submerged aerator to enhance oxygenation in the vicinity of the plant roots and associated 651 microorganisms, thus increasing the nutrient degradation process (Lu et al., 2015) (Figure  652 12.6).While these hybrid systems appear to enhance pollutant removal from the water column 653 compared to their macrophyte-only counterparts (Guo et al., 2014;Li et al., 2010), the added 654 complexity may impact on the utility of FTW as a phytoremediation system. With increasing 655 complexity of FTW design there is an increase in pollutant removal efficiency, cost and maintenance, 656 but a decrease in user uptake given the added management of submerged plants or solar PV 657 systems. A focus on maximising removal efficiency over the simplicity of the system may create 658 barriers for uptake by stakeholders such as farmers, land managers and government organisations 659 looking for low-cost low maintenance treatment options, especially within developing countries. A 660 useful exercise might be to compare the economics, maintenance requirements and user experience 661 of hybrid versus conventional FTWs to determine when increasing FTW complexity is appropriate. 662 The coverage of FTW over the target water body is also important, as indicated by a meta-663 analysis showing that vegetation cover is significantly correlated with the removal of NH4 - (Pavlineri 664 et al., 2017). Although increasing FTW coverage reduces atmospheric diffusion, oxygen is supplied to 665 water by emergent plants via root oxygenation (Xiao et al., 2016;Yeh et al., 2015). Furthermore, in 666 eutrophic waters this coverage may inhibit algal primary productivity, which may be beneficial for 667 mitigating the potential for occurrences of large algal blooms . The optimal 668 coverage of FTWs has been reported as 10-25% (Marimon et al., 2013), although generally there is 669 wide variation in the literature with values of between 100 %, 50 % and 5-8 % being reported as 670 acceptable for water treatment (Pavlineri et al., 2017). McAndrew & Ahn (2017) also note that 671 hydraulic retention time and plant productivity are important for determining removal efficiency. 672 Surface cover therefore needs to be considered in tandem with hydrology and macrophyte 673 selection. As the focus within the literature is on coverage, there has been no clear attempt to look 674 at the different surface arrangements of FTW on the water surface. For example, targeting of an 675 area, such as water inlet or outlet to a lake may be more beneficial than increased FTW coverage 676 over the target water body. Clearly, the coverage and area of FTW treatment is context-specific but 677 there is likely to be significant potential in investigating spatially targeted phytoremediation. 678 Finally, the poor design and management of FTWs is a topic that is rarely discussed within the 679 literature. FTWs have the potential to be pollutant sources should the biomass not be continually 680 harvested and removed, or if water birds attracted to the FTWs defecate into the water inputting 681 nutrients and microbial contaminants (guanotrophication). Nutrient-rich growth media such as peat 682 may also leach nutrients into the target water body compared to more inert coir fibre (Lynch et al., 683 2015). The placement of FTWs in watercourses must also be givenfull consideration aswater birds 684 and recreational users may also use the target waterbody. FTWs potentially slow the velocity of 685 water in small water bodies such as ditches, which may conflict with farming interests where good 686 drainage is required. As with any good catchment management practice, appropriate consultation 687 with stakeholders is important for success. are harvested for effective removalof heavy metal and organic pollutants from the planting system. 695 Allometry of pollutants within plants varies according to species, but is also influenced by the 696 environmental conditions in terms of nutrient availability (Barrat-Segretain, 2001;Demars and 697 Edwards, 2007). 698

Typha domingensis, Eichhornia crassipes, Pistia stratiotes and Myriophyllum 699
aquaticumpreferentially storeN and P in the shoot compared to therootsorrhizome (Table 12.6), 700 although nutrients can be translocated through the plants leading to temporal dynamismin element 701 distribution driven byplant phenology and diurnal metabolism (Masclaux-Daubresse et al., 2010;702 Hawkesford et al., 2011;Eid et al., 2012).More than 50% of N can be storedinbelow-groundplant 703 parts by the end of a growing season (Vymazal, 2007).Phragmites australisgrownin either natural 704 waters or a waste water infiltration pond demonstrated a clear seasonal pattern in the translocation 705 of nutrients from above-groundto below-ground parts as the end of the growing season 706 approached (Meuleman et al., 2002). Early in the growing season N and P concentrations are higher 707 due to sink demand during active growth before concentrations decrease gradually through the 708 season as plants begin to senesce. 709 Coinciding with the decrease in nutrient concentrations in above-ground biomass, below-710 ground concentrations of N and P increase, representing the preparation for plant senescence with 711 nutrient storage in the roots and rhizomes for the following season's growth (Garver et al., 1988). 712 Meuleman et al. (2002) suggested that harvesting during the winter meant that only 9% of N and 6% 713 of P associated with nutrient loading was removed, whereas, harvesting above-ground parts during 714 peak nutrient storage in summer enhanced removal to 40-50% of N and P. Seasonality is 715 important,although seasonal effects will differ between temperate, subtropical and tropical zones 716 with macrophytes in the latter two zones showing less element translocation and therefore enabling 717 multiple annual harvests (Vymazal, 2007).Macrophytes may perform poorly if nutrient translocation 718 to the rhizome is inhibited by harvesting during the active growing period (Tanaka et al., 2017), 719 although the issue of nutrient allocation is less problematic for floating macrophytes and emergent 720 macrophytes deployed in FTWs as the full plant can then be harvested . 721 Studies on element allocation tend to reportabsolute concentrationsto determine if a 722 species is a betterabove-ground or below-ground accumulator. The potential for pollutant uptake 723 and removal by harvesting the areal parts is a function of both concentration and the biomass 724 produced (Polomski et al., 2009). For example, although shoot concentration of N in Pistia 725 stratiotes(13.93mg/g) was greater than inEichhornia crassipes(10.16mg/g) in a study of nutrient 726 recovery, the total areal shoot storage of N for Eichhornia crassipes was over four times higher due 727 to its greater biomass (Polomski et al., 2009). This demonstrates that it is more effective to harvest 728 plants with greater above-ground biomass andmoderate tissue concentrations of the pollutant of 729 interest, rather target plants with lower biomass but higher tissue concentrations (Duman et al., 730 2007;Vymazal, 2016). 731 In eutrophic waters light is commonly the limiting factor for growth and plants therefore tend to 732 allocate nutrients to above-ground growth to maintain efficient light capture, while excessive 733 nutrient availability negates the requirement for belowground storage (Polomski et al., 2009;Lynch 734 et al., 2015); this also maintain intra-specific competitive advantages in these environments and can 735 be exploited as part of a phytoremediation management strategy . Where non-hyperaccumulator 736 plants are grown in a substrate where high concentrations of heavy metals and organic pollutants 737 are present, physiological mechanisms within these plants often limit the transport of these 738 compounds to above-ground tissue to mitigate damage to important cells, such as those responsible 739 for photosynthesis Verkleij et al. 2009). 740 Thepreference for below-ground storage by emergent macrophytes has been demonstrated in 741 multiple studies, as listed in Table 12.6. However, there are some occasions where metals are found 742 at greater concentration in aerial parts, such as Pb in Cyperus esculentus, Zn in Glyceria maxima, Mn 743 in Phragmites australis and Cu in Phragmites australis (Table 12.6), which suggests that specifically 744 classing species as above-ground or below-ground accumulators of specific pollutants may be 745 inappropriate. Furthermore, not all studies capture the full seasonal dynamics of nutrient or 746 pollutant translocation and allometry under different concentration regimes, and therefore, to 747 enable sound recommendations on harvesting during phytoremediation projects, further studies to 748 characterise chemical allocation over time of key species should be carried out to ensure pollutant 749 removal is appropriatelytargeted. 750

752
There is debate within the phytoremediation literature as to the relative importance of 753 macrophytes in removing pollutants compared to the independent microbial degradation. This 754 perspective primarily comes from observations showing that unplanted CWs can match or 755 outperform planted CWs in terms of pollutant removal (Cardinal et al., 2014). In addition to 756 microbial activity, processes such as sedimentation in P stabilisation and removal, and the 757 photodegradation of PPCPshave also been noted as important (Cardinal et al., 2014;Tanner & 758 Headley, 2011;. Microbial activity is also an important factor for 759 enablingphytodegradation of pollutants, however, the independent role of microbial communitiesis 760 now receiving much more attention (Houda et al., 2014). Improved understanding of how microbial 761 activity contributes to pollutant degradationis essential because it not only influences removal rates 762 but may have implications for the value of harvesting plant biomass and post-remediation resource 763 recovery if the actual plant uptake and sequestration (phytoextraction) of target pollutants is low. 764 There is an abundance of microorganisms associated with macrophyte roots that influence 765 the removal and degradation of pollutants (Stottmeister et al., 2003;Faulwetter et al., 2009). These 766 include bacteria that assist in nitrification and denitrification for the transformation and removal of 767 excess N, and biological mineralization of organic P (Valipour and Ahn, 2016). These processes are 768 integral to the efficient functioning of CWs but the role of macrophytes in facilitating and enhancing 769 the metabolic processes of these microorganisms is still not well understood, although it is likely 770 thatthe rhizosphere provides an energy source for microorganisms (Thijs et al., 2016). Redox state, 771 dissolved oxygen content and temperature are common limiting factors for different 772 microorganisms (Truu et al., 2009), and the potential for macrophytes to oxygenate the substrate 773 surrounding their below-ground organs can alsofacilitate the growth of microbes in the rhizosphere 774 (Pavlineri et al., 2017). 775 CWs are highly engineered, with multiple design elements that may influence the 776 abundance and diversity of microorganisms.Consequently carefully designed experiments are 777 required to explore the potential role of theplant microbiome in phytoremediation. Applying this 778 knowledge is particularly important for developing novel environmental engineering solutions such 779 as FTWs. The formation of microbial biofilms on the underside of FTWs and plant roots has been 780 suggested as a key removal pathway for nutrients and heavy metals (Tanner et al., 2011). Wang & 781 Sample (2014) found that unplanted FTWs had similar removal efficiencies compared to those 782 planted with monocultures of Pontederia cordataand Schoenoplectus tabernaemontani (Figure 12.7). 783 In this study, and elsewhere, temperature was a key factor in the performance of FTW which has 784 beenrelated to changes in microbial activity (Van de Moortel, 2011;Wang & Sample, 2014b). In 785 contrast,  were unable to link microbial community traits associated with FTWs 786 biofilm such as ribotype number and diversity index to the removal efficiency of pollutants. 787 Given the conflicting evidence on the relative importance of plants and biofilms in 788 phytoremediation,a 'meta-organism' approach to phytoremediation is now required to appreciate 789 the multitude of factors and process at work (Thijs et al., 2016;Feng et al., 2017). Further studies are 790 required in these areas that employ suitable control treatments, along with adequate spatial and 791 temporal characterisation of microbial communities for different macrophytes in monoculture and 792 polyculture, and growth media. Furthermore, within these studies the mass balance of pollutant 793 allocation should be investigated to fully assess where and how pollutants are being stored and 794 translocated. Radio-labelled isotopes have been successfully employed to quantify cycling of 795 nutrients within CWs (Truu et al., 2009). However, such techniques have not been employed during 796 FTW studies, where the application of radio-labelled isotopes would provide an opportunity to 797 understand the biochemical cycling with these novel systems. Finally, after adequate 798 characterisation of microbial communities and their relation to the plant and associated abiotic 799 environment, there may be new opportunities to enhance the microbial community to promote 800 pollutant removal (Glick, 2003;Thijs et al., 2016). The process of phytoremediation has primarily been concerned with maximising the 805 efficiency of water treatment,whilst the benefits of phytoremediation over and above remediation 806 have essentially been overlooked. Clearly, water treatment is the primary ecosystem service in the 807 provision of safe and clean water; however, the planting of vegetation within the environment 808 creates new habitats fororganisms . For example, the presence of artificial floating 809 islands improved chickproductivity ofBlack-throated Divers (Gavia arctica) by 44 % in waterbodies 810 with these structures (Hancock, 2000), indicating a potential combined role for FTWs in water 811 treatment and improved habitat connectivity. Similarly, a 15year project investigating the 812 environmental benefits of creating treatment wetlands to ameliorate mine tailing effluentsfound 813 that there was a high abundance and diversity of protozoa, higher plants, terrestrial animals, and 814 birds (Yang et al., 2006). 815 In addition to habitat provisioning there is also the potential for facilitating pollination and 816 carbon sequestration (Nesshöver et al., 2017). The capacity for the latter may depend on the post-817 remediation stage and the reuse of the biomass. Cultural services can also be provided by an 818 improvement in the aesthetic appeal of an area with increased vegetation (Masi et al., 2017). This is 819 most likely in urban waterways where FTW might provide attractive green infrastructure (Olguín et 820 al., 2017). There is a need to quantify and assess ecosystem services associated with 821 phytoremediation projects in order tobetter appreciate the multiple benefits generated from this 822 form of water treatment. 823 12.10.2 Resource recovery 824 825 The potential to generate large volumes of biomass through phytoremediation means that there 826 are opportunities for resource recovery within the process (Gomes, 2012). Post-remediation 827 biomass re-use streams (PBRSs) are the disposal process and utilisationof the harvested plant tissues 828 of macrophytes used for phytoremediation (Gomes, 2012).As macrophytes are able to remove and 829 assimilate metals there is certainly potential for the recovery of metals such as gold, Cu and Ni 830 (phytomining) (Anderson et al. 2005). To date, most research in this area has focused on terrestrial 831 plants and soils contaminated through industrial mining (Rosenkranz et al., 2017). However, there 832 may be potential to explore metal-contaminated waters and sediments of wetlands used to treat 833 mine-tailing effluents. The usefulness of this process depends on the current market value of target 834 metals and the economic benefits associated with this form of phytoremediation (Sheoran et al., 835 2009). 836 The use of macrophytesas biofuels is another possibility and is a feasible option to increase the 837 value of phytoremediation if there is a market for biomass. An economic assessment by Jiang et al. 838 (2015) found that high biomass production plants are required to make this a profitable venture. 839 However, different options need to be considered in pre-treatment, such as de-wetting and 840 briquetting,since fresh plant biomass comprises up to 90% water (Newete and Byrne, 841 2016).Macrophyte biomass may also be used for animal feed, or to make compost or 842 biochar (Quilliam et al., 2015;Tanaka et al. 2017). Quilliam et al. (2015) discussed in detail the issues 843 with these PBRSs in terms of the transfer of pathogens, bio-magnification of heavy metals and 844 propagation of invasive species. A phytoremediation decision-making system that couples the target 845 pollutants and the PBRS would allow the resource recovery options to be established early in the 846 process (Song and Park, 2017). For example, the remediation of a eutrophic lake would seemto link 847 well with composting or animal feed PBRS given the potential for high nutritional content.However, 848 if heavy metal or pesticide contamination also is identified, then a biofuel or phytomining PBRS may 849 be more appropriate. Larger scale pilot tests of aquatic phytoremediation are required, and these 850 should explore the feasibility of using produced biomass in PBRSs. 851

853
This chapter has outlined the potential of aquatic phytoremediation to provide efficient, 854 multi-targeted and sustainable remediation solutions for polluted waters. A summary of a proposed 855 research agenda required to fulfil the potential of these systems is presented in Table 12.7. Given 856 the wide range of organic, inorganic and biological pollutants that can impact surface waters there is 857 a need to steer phytoremediation towards a context-specific approach that allows the remediation 858 of multiple water body types, and waters affected by a range of pollutants. 859 With the development of novel ways to deploy macrophytes, such as by FTWs, there are 860 emerging options for spatial flexibility of applying phytoremediation, which are relatively 861 inexpensive. Larger scale pilot studies are required in this respect to assess the realistic 862 opportunities for use. At present there are a wide range of macrophytes of different growth forms 863 that have been established as efficient accumulators of pollutants. A further focus is required to 864 investigate the remediation potential of submerged species and to establish new accumulators that 865 may be used. Importantly, some of the key hyperaccumulators are considered invasive and would be 866 unsuitable to be deployed in natural surface waters. A proposed advancement for phytoremediation 867 systems is to consider the benefits of a plant community based-approach that assembles 868 polycultures of macrophyteswith good accumulation capacity for different pollutants, enabling 869 multi-targeted remediation. Here, the need for a logical system of macrophyte selection based on 870 plant removal efficiencies and environmental tolerances, and target pollutant specifications, 871 requires development. 872 The process of macrophyte phytoremediation still requires a deeper understanding of how 873 to enhance removal efficiency and ensure sustainable harvesting of macrophytes. Understanding the 874 spatial and temporal dynamics of pollutant translocation within macrophytes is crucial for 875 permanent pollutant removal from water and for maintaining the economic value of different PBRSs. 876 Furthermore, a 'meta-organism' approach needs to be considered in future phytoremediation 877 studies to establish the role of plant-associated microbial communities. There may be untapped 878 potential in manipulating these microbial communities for enhanced performance. 879 Finally, the focus of phytoremediation has been on the water treatment aspect, whilst there 880 is growing recognition of the capacity of these ecological engineering strategies to provide 881 ecosystem services such as carbon sequestration and biodiversity support. Thesebenefits need to be 882 better quantified to determine the added-value of phytoremediation. With the waste management 883 sector shifting towards a life-cycle approach, there are clear opportunities for resource recovery 884 through identifying PBRSs such as composting, biofuel production and animal feed. These PBRSs 885 require further exploration in terms of their safety, value and ability to link directly with the target 886 pollutants removed (Figure 12.8). A life-cycle approach needs to embedded in prospective aquatic 887 phytoremediation projects,to ensure that target pollutant(s) are being considered in tandem with 888 the PBRS, whilst the frequency of harvest and replacement/regrowth of macrophytes is properly 889 linked into the remediation of the target pollutant (Figure 12.8).       (2014) Table 12.7: Summary of the aquatic phytoremediation research agenda required to deliver efficient, multi-targeted and suitable phytoremediation. Research areas, specific lines of investigation and their priority are highlighted.

Identify new macrophyte accumulators for emerging pollutants
To what extent can macrophytes assimilate and degrade PPCPs and pathogens?

Plant community-based remediation
Evaluate potential for multi-targeted remediation in plant polyculture incorporating temporal/phonological differences and asses plant competitive effects