Field and mesocosm test methods to assess the performance of biodegradable plastic 1 under marine conditions 2 3

16 The pollution of the natural environment, especially the world’s oceans, with 17 conventional plastic is of major concern. Biodegradable plastics are an emerging 18 market bringing along potential chances and risks. The fate of these materials in the 19 environment and their possible effects on organisms and ecosystems has rarely been 20 studied systematically and is not well understood. For the marine environment, reliable 21 field test methods and standards for assessing and certifying biodegradation are 22 lacking. In this work we present newly developed field tests to assess the performance 23 of biodegradable plastics under natural marine conditions. These methods were 24 successfully applied and validated in three coastal habitats (eulittoral, benthic and 25 Manuscript


pelagic) and in two climate zones (Mediterranean Sea and tropical Southeast Asia). 26
Additionally, a stand-alone mesocosm test system which integrated all three habitats 27 in one technical system at 400-L scale independent from running seawater is 28 presented as a methodological bridge. Films of the positive control test material 29 polyhydroxyalkanoate copolymer (PHA) and the negative control low density 30 polyethylene (LD-PE) were used to validate the systems. While LD-PE remained intact, 31 PHA disintegrated with speed depending on the habitat and the climate zone. Together 32 with the existing laboratory standard test methods, the field and mesocosm test 33 systems presented in this work provide a 3-tier testing scheme for the reliable 34 Introduction 38 Biodegradable plastic materials are being introduced into the market with a growing 39 share in recent years [1], with common uses including food packaging, very lightweight 40 bags and agricultural applications. The possible environmental benefit of 41 biodegradable plastics is dependent on the application in which they are used. For 42 example, certified soil-biodegradable mulch films [2] are utilized to substitute 43 conventional plastic polyethylene films, especially for those applications in which 44 complete recollection of the film is not possible. After use, certified soil-biodegradable 45 mulch films are plowed into the soil and biodegraded by microbes, avoiding the 46 accumulation of non-degradable plastic fragments in the fields. It is fundamental to 47 keep in mind that biodegradation is the remineralization of the material to CO2 (and/or 48 CH4 under anoxic conditions) and water, and the conversion to biomass by the 49 metabolic action of microbes. The proof of environmental biodegradability, i.e. beyond controlled lab test conditions, is the first prerequisite for a polymer to be considered as 51 a sustainable alternative to non-biodegradable materials in a specific environment. 52 Furthermore, biodegradable substitute materials are opted for as a mitigation against 53 marine plastic pollution (e.g. [3]). However, while there exist certification schemes for 54 industrial compostability (e.g., based on standards [4][5][6]) or soil biodegradability [2], 55 there is a lack of standards and methods for assessing marine biodegradability. It is 56 well recognized that positive experimental results on biodegradation of polymers in 57 one environment (e.g. in industrial compost) do not automatically imply comparable or 58 sufficient biodegradation rates of that polymer in another system (e.g. in the marine 59 environment) [7][8]. 60 In fact, the knowledge on the biodegradability of polymers in freshwater and marine 61 environments is still very limited and experimental data vary strongly. Hence, there is 62 a clear need (and also clear international political call, e.g. The European Green Deal, 63 12.11.2019 [9]) for reliable environmentally relevant standardized test methodologies 64 to produce comparative and scientifically valid results and for the verification of claims 65 based on sound knowledge rather than assumptions [10]. Data from such tests will 66 help to assess environmental benefits and potential risks and will allow for their 67 comprehensive life cycle assessment. 68 The number of tests on biodegradable plastics that have been conducted directly in 69 dimensions leaving a surface of 320 cm 2 of material directly exposed. The frames, the 152 mesh and the test material were then assembled with plastic bolts and nuts (Nylon 6.6,153 www.kunststoffschraube.de, Singen, Germany) and each test frame was individually 154 tagged with a code made from cable markers (HellermannTyton, Tornesch, Germany) 155 fixed with a cable tie. The dimensions of the test items were chosen to provide enough 156 material for subsampling (Fig 1, left) for further analyses. where (conventional) plastic litter is found in the ocean [33,35]. Italy (42°44'06.5"N 010°10'33.5"E) and was used to wash the sediment and to fill the 184 mesocosms. Natural marine sediment of siliclastic origin for the eulittoral (field and 185 mesocosm) tests was retrieved at about 0.1 m water depth from the beach of Fetovaia, 186 Isola d'Elba, Italy, (42°44'00.1"N 010°09'15.3"E) and is called "beach sediment". with the open sea by means of several inlets. There is a small irrigation trench that 200 occasionally brings some freshwater runoff to the basin. The ecological character of the site is that of a typical Mediterranean coastal lagoon with weak estuarine influence. 202 Beach sediment was prepared for the eulittoral tests as described above. 203 The pelagic and benthic tests were performed from June 2014 to October 2016 in the 204 marine protected area of the National Park Tuscan Archipelago off the island of 205 Pianosa (42°34 '41.4"N 010°06'30.6"E, Fig 3). The test system was set up 206 approximately 500 m from the south-eastern shore close to Punta Secca. The 207 ecological character of this site is that of a typical Mediterranean coastal habitat with 208 low anthropogenic impact and the risk of loss or disturbance by human activities was 209 regarded as very low. 210 211 Eulittoral field test system 212 This test system mimicked an intertidal sandy beach in which plastic was buried and 213 experienced changing conditions of being wetted and falling dry with the fluctuations 214 of the sea level caused by tides and waves (Fig 4). A plastic (PP) bin of approx. 60 L 215 volume was used to confine the sediment and the samples. The bottom of the bin was 216 perforated with holes of 25 mm diameter to allow the ambient water to infiltrate. Two 217 layers of plastic mesh (PVC-covered glass fiber 1 x 1 mm and polyester gauze 280 x 218 280 m were placed above the holes to prevent the loss of sediment. A layer of 17 cm 219 of beach sand was filled into the bin and a plastic pipe of 32 mm diameter, covered by 220 fine mesh at the top opening, was installed as a short-cut for the rising and falling water 221 to facilitate the drainage of the sediment. Three test frames were buried at a sediment 222 depth of 10 cm with an inclination of about 11° from horizontal to allow the overlaying 223 water to run off after each flood event. The filled bins were placed in groups of five on 224 a wooden rack in order to position the test specimens in the mid-water line in a coastal 225 lagoon. The rack structures and bins were covered by wire barbs to prevent 226 disturbance from birds. 227

229
(1) wire barbs to deter birds, (2) 60-L plastic bin, (3) sand from a local beach, (4) test item frame, (5) 230 fine mesh (280 µm), (6) coarse mesh (1000 µm), (7) perforations in bottom of the bin, (8) equilibration 231 pipe, (9) fine mesh (280 µm) on the top of the pipe to allow for seawater outflow at high tide. Top center: 232 3 specimens in the test bin before covering with an additional layer of sand. was neutrally buoyant and was in contact with only seawater as a matrix (Fig 5). The 241 test frames were fixed with cable ties perpendicularly on top of each other to a plastic 242 bar, four of which were mounted to a plastic cylinder Technoplast,243 Lahnstein, Germany) with stainless steel bolts and nuts to form 4 radial wings. The 244 cylinder was suspended upright at a water depth of about 20 m from a float with a lift 245 of approximately 5 L and kept in place by a concrete anchor weight (50 kg) at the 246 seafloor; this anchor was connected by a stainless steel cable with stainless steel bolts 247 and nuts, or shackles. To prevent corrosion the steel cables were equipped with zinc 248 anodes, fixed with stainless steel bolts and nuts. 249 contact with seawater and sediment as matrices (Fig 6). 3 x 5 test specimens were 259 mounted on a rack of plastic bars and placed on the seafloor at approximately 40 3). This location was chosen because it has been used for generations by local 275 people to permanently anchoring floating fishing huts (therefore a historical proof of 276 good storm protection). The bay is also very narrow and not suited for fishing methods 277 that might have interfered with the test systems such as purse seining or trawling. The 278 benthic system was deployed at a water depth of 32 m on sand to avoid the coral reef. 279 The pelagic test was suspended in 20 m water depth. The ecological character of the 280 site is that of a typical coastal area in the wet tropics, with high water temperature all 281 year round and elevated nutrient content due to terrestrial run-off.
As a modification to the experiments in the Mediterranean Sea the mesh used for all 283 test frames was 2 x 2 mm (Fig 1). There were no eulittoral tests done. The mesocosm experiments simulated, in a multi-100-liter tank under controlled 288 conditions beyond laboratory flask scale (usually a few 100 mL of volume), the 289 degradation of the polymers PHA and LDPE in the marine environment in three coastal 290 habitats: eulittoral, pelagic and benthic (Fig 7). These habitats were mimicked in a 291 combined tank system as follows: PE-HD plastic tanks (Dolav GmbH, Bad Salzuflen) 292 with inner dimensions 93 x 113 x 60 cm were set up in triplicates in a climate room at 293 21 °C. Each set consisted of the eulittoral test tank placed on top of the tank with the 294 benthic/pelagic test system, connected by a closed-system circuit of about 400 L 295 seawater. Water was pumped into the upper tank in a way that simulated a semidiurnal 296 tide, creating complete flooding every 12 h and a complete draining 6 h later. Tests 297 were conducted twice for about 10 months each and termed year 1 (y1) and year (y2), 298 with 4 sampling time intervals each. Before repeating the test, the matrices water and 299 sediment were renewed. 300 301 Eulittoral mesocosm test system 302 The top tank (Fig 7) mimicked the eulittoral (intertidal) scenario and was filled with a 303 layer of 15 cm of beach sediment. The incoming water was led by a hose to a 304 perforated pipe (diam. 12 mm, L = 1 m) which was laid onto the sediment surface. The 305 water inlet was connected to the benthic/pelagic test tank below by a EHEIM compact 306 had an outlet in the center of the bottom (diam. 50 mm) which was covered by gauze 308   The physical sediment parameters grain size distribution, permeability and porosity 358 were analysed with standard methods (e.g. [36]) and are given in Table 1.

399
The measured water temperature was 20.5±1 °C (mean±standard deviation) (set: 21 400 °C) and mean light intensity on the sediment surface of the benthic tests was 11.56 401 μmol photons·m -2 ·s -1 . Salinity was about 39 with a variation of ±1 for both years (set: 402 38.5). The pH was stable at 8.1±0.1 and the oxygen concentration was close to air 403 saturation (98±2 %). Water chemistry: Throughout the experiments, water in all tanks 404 and porewater from the eulittoral sand contained low to moderate levels of nutrient 405 related parameters (TN, NO2, NO3, NH4, TP, TOC, DOC, Chl a, Phaeophytin, etc.; S1 406 Table). Compared to the field data nutrient concentrations were similar or slightly the toxic substances as heavy metals, organotin compounds and POPs were detected. 409 Sedimentology and sediment chemistry: The beach sand of the eulittoral tests was 410 mainly of siliclastic origin and was classified using [37] as medium sand. The sediment 411 used for the benthic experiments was composed mainly of carbonate minerals and 412 characterized as fine sand in both years. Accordingly, porosity and permeability were 413 slightly lower for both sands in the second year (Table 1) whereas nutrient contents 414 were similar and low (S2 Table). Metal concentrations were low or below detection 415 limit. None of the known anthropogenic toxins analysed in the sediments used for the 416 experiments were detected (S4 Table). between 20 and 80 %, and rarely reached 100 % air saturation (see also Table 2). 429 Water chemistry: The porewater in the bins throughout the experiments had low to 430 moderate levels of nutrient related parameters, however compared to the benthic and 431 pelagic test sites they were slightly elevated (S1 Table). Dissolved iron, manganese sand used in the test bins was fine (mean 214±5 µm yr1) and medium (mean 278±5 435 µm yr2) sand. The porosity and permeability were therefore higher in the second year 436 (Table 1). Nutrient concentrations were low or b.d.l. and of the metals analysed only 437 zinc, nickel and chromium were detected in low concentrations. Organotin compounds, 438 pesticides and their metabolites could not be detected in the sediment (S4 Table). through the water column. At the benthic test autumn fluctuations were more 451 pronounced due to occasional storms that led to a mixing of the water column. At the 452 depth of the pelagic test system (20 m) solar irradiation was about 13 % and at the 453 benthic test system (40 m) about 3 % of the surface level. Salinity was at 38, pH was 454 8.2 and oxygen content was around 100 % air saturation at both sites. 455 Pelagic and benthic water chemistry: The water at both test sites as well as the 456 porewater at the benthic test site had low levels of nutrient related parameters 457 throughout the experiments (S1 Table). Dissolved aluminium could be repeatedly 458 detected at low concentrations, as well as iron and manganese in the second year. 459 Benthic sedimentology and sediment chemistry: The benthic sand was fine sand 461 (mean 213±26 µm) with low concentrations of nutrients and some metals (iron, 462 manganese, lead, chromium) (S2 Table). Organotin compounds and pesticides and 463 their metabolites could not be detected. No toxins or POPs were detected in the 464 sediment (S4 Table).

481
Pelagic and benthic water chemistry: The water at both test sites as well as the 482 porewater at the benthic test site had low levels of most nutrient related parameters 483 throughout the experiments (S1 Table). The silica, chlorophyll a and phaeophytin 484 concentrations however were higher than at the tests sites in the Mediterranean Sea. 485 Metal concentrations were all b.d.l., only cadmium was detected once in very low 486 concentration (S1 Table). No toxic substances or POPs were detected (S1 Table). 487 Benthic sedimentology and sediment chemistry: The benthic sand was fine sand 488 (mean 197±47 µm) with low nutrients and low concentrations of some metals (iron, 489 manganese, lead, chromium; S2 Table). No toxins or POPs were detected in the 490 sediment (S4 Table). benthic) or dug out of the sediment (eulittoral), gently rinsed in ambient water, packed 495 camera (Canon EOS 5DMkII, or a SONY A7 III, with a 100 mm or 50 mm macro lens). 499 Subsamples were taken for later optional microscopic and molecular analyses, which 500 are not subject of this publication. The remaining sample was rinsed in deionized 501 water, air dried and archived. 502 The disintegration of each specimen was determined photogrammetrically [13]. Dried 503 samples from the Mediterranean Sea experiments were scanned on a flatbed scanner 504 (LIDE 210, Canon Inc.). The samples from the SE Asia experiments were 505 photographed immediately after retrieval. The images thus obtained were analyzed for 506 area loss using ImageJ software https://imagej.nih.gov/ij/) or GIMP 507 (http://www.gimp.org/) by outlining the boundaries of the remaining test material that 508 had been exposed and not covered by the frame. Subsequently, the ratio of total area 509 vs. lost area was calculated as % area loss. For LDPE, disintegration was not detected in any of the tests, and will not be described 529 further in the results section. PHA disintegration was observed in all tests (Fig 12). replicates were disintegrated to over 65 %. In the y2 experiment PHA disintegration 536 was slower and reached just over 30 % after 10 months for one replicate. All replicates 537 were disintegrated to more than 10 % after 5 months. Generally, the disintegration of 538 all replicates of all sampling intervals was heterogeneous. 539 540 Benthic test (sublittoral seafloor scenario) 541 PHA showed disintegration, which however differed by year (Fig 12, top row, left). One 542 sample showed disintegration of 31 % after 5 months and two samples 60 -75 % after 543 22 months. The disintegration of all other samples in both years was below 5 %. In the 544 y2 experiment PHA disintegration was slower and only one sample reached more than 545 6 %. Generally, the disintegration of all replicates was heterogeneous. 546 At sampling, all specimens were still intact after the exposure time of max. 22 months 549 (Fig 12,top row,right). In some specimens a thinning of the material was observed, 550 visible by more translucent areas. Small holes were detected in only a few samples, 551 but accounted for less than 1 % area loss. bottom row). A conspicuous longitudinal crack was observed in all specimens after 559 exposure, in the cases where there was still material left (Fig 9). This might have been 560 be a consequence of material shrinking when exposed to water for a longer period of 561 time. The crack was not taken into account as loss by disintegration but ignored for the 562 area measurement.

570
Disintegration was observed in all PHA samples, however the accurate measurement 571 of the lost area was not possible due to heavy fouling of the support structure and the 572 polymer itself. Also, the layer of the fouling organisms (mainly encrusting red algae, 573 sponges, hydrozoans and bryozoans) often was bonding the mesh and the polymer together. Splitting open the test frame during post-sampling in the lab caused further 575 fragmentation and thus a mechanical destruction of the materials as an artefact. 576 Another factor of sample deterioration in the pelagic tests was linked to animals found 577 in between the meshes. Presumably, crabs and clams had settled as larvae on the test 578 material and after having grown above mesh size they had been trapped between the 579 two layers of protecting mesh. Traces of their movements point to these animals having 580 mechanically destroyed the polymer (Fig 10,

608
The disintegration was gradually progressed with time, with most samples 609 disintegrating less than 50% after 308 (y1), or 270 (y2) days of exposure. The rate of 610 disintegration differed between habitats and was heterogeneous between replicate 611 samples of one sampling interval. PHA disintegration was fastest in the benthic tests 612 (Fig 12, center row, left), and slowest in the y1 experiment in the eulittoral tests (Fig  613   12, center row, center, empty circles), but in y2 slowest in the pelagic tests (Fig 12,  614 center row, right, filled circles). 619 experiment were kept exposed over the duration of the y2 experiment for a total of 678 days. Top row, 620 center: eulittoral field test MED: One set of replicates from the y1 experiment exposed over the duration 621 of the y2 (686 d). Top row, right: pelagic field test MED: Disintegration in all samples was less than 1%.

622
Two replicates from the y1 experiment exposed over y2 (678 d

Additional observations
Several measures were taken to prevent negative impact by animals on the test 629 system. However, some disturbances resulted from the interaction of (macro-) 630 organisms with the exposed materials in the field, and to a lower extent also in the 631

mesocosms. 632
Eulittoral field tests: The picking of birds in the eulittoral test bins was successfully 633 prevented by wire barbs on top of the test system. Infauna like worms or clams were 634 not observed in the eulittoral test bins. A structural problem for the racks made from 635 untreated larch wood resulted from the activity of wood-boring bivalves ("shipworms") 636 that had destroyed the posts after two years to the brink of collapse (Fig 13). cables were chosen over plastic rope because turtles and some fish are known to 652 snatch ropes while feeding on sponges and other encrusting organisms. However, on 653 several occasions bite marks on the frames and mesh were observed, but a sample 654 was never found bitten all the way to the specimen. On two occasions cable ties that served to fix the codes to the samples were found cut open, and once a large puffer 656 fish was observed grazing on the pelagic test system, having just bitten off a code tag.  The mesocosm tests also were proven practical and generally suited to run for a longer 705 time period in a closed circuit. The salinity could be easily adjusted, and the pH was stable over the course of the experiment. In the pelagic and benthic tests however, 707 photosynthetic biofilms may complicate the experiments by causing a patchy 708 overgrowth on the samples. Being of a limited amount of water and sediment, the biotic 709 community originating from an inoculum of random diversity can lead to the dominance 710 of few organisms (cyanobacteria and/or algae) covering large areas in each of the 711 mesocosm tanks. In order to avoid these differences, also between tanks, we 712 recommend conducting the experiments in the dark. Also, higher but still 713 physiologically relevant temperatures for the respective climate zone (e.g. 25°C in the 714 Mediterranean tests) could be used to accelerate the processes and render the tests 715 more efficient. The heterogeneity of the disintegration between replicate samples is 716 not to be seen as an artefact of the mesocosm tank tests but was also observed 717 between samples in the field tests (Fig 12). This is interpreted as a reflection of the 718 patchiness of the microbial community according to small-scale changes of 719 environmental conditions, as similarly observed also for example, in marine coastal 720 sands (e.g. [38]). 721 722 Using disintegration as a proxy for biodegradation only after intrinsic 723 biodegradability has been proven in lab tests 724 In an open-system biodegradation test, where metabolites like CO2 or O2 cannot be 725 tracked directly with commonly available (i.e. not labelled) plastic materials, degree of 726 disintegration is a feasible parameter to be measured in a simple and cost efficient 727 manner. The disintegration can however only be considered a valid proxy under the 728 assumption that: 1) the physical forces are minimized or excluded in the open system 729 tests and, more importantly, that the test material has shown to be inherently 730 biodegradable in a (standard) lab test (water: [23,28,29]; sediment-water interface: PHA Mirel P5001 that was also used in lab tests during the Open-Bio project. PHAs 733 were also proven by others to fully biodegrade (PHB by [15] in seawater according to 734 [23]; PHA Mirel P5001 by [40] in seawater according [23]; PHBV by [41] in seawater 735 with biofilm from fish tank as inoculum). 736 As we were testing thin polymer film, we chose to measure area loss by 737 photogrammetry instead of weight attrition [32] to assess the disintegration of the test 738 Degradation under the frame differs from exposed area of interest 773 Due to the design of the HYDRA test frames, parts of the test material which were not 774 intended as area of interest were covered. Only the material exposed to the matrices 775 was analyzed as described. However, the shielding of some of the material by the 776 frames brought along an unintentional co-testing of the material under presumably 777 oxygen-reduced conditions and gave hints for the anaerobic degradation of 778 biodegradable plastic. Since some plastics may degrade faster under anoxic 779 conditions, this observation should be addressed by a specific study. 780 The specimens exposed in the pelagic test system in SE Asia were quickly covered 783 through heavy fouling. All the materials such as frames, mesh and support structures 784 were already overgrown after a few months by e.g. coralline algae and sponges which 785 could not be removed without strong physical impact on the test material -a factor that 786 had been a priority to avoid in the experimental design. Due to rather large sampling 787 intervals, the rapid succession of the organism community and the general impact of 788 fouling could not be studied. A small-scale observation with regular samplings every 789 few days to weeks would be required for a good temporal resolution to understand 790 whether fouling would hinder or accelerate polymer degradation. A previous study in 791 the Mediterranean Sea indicated a decrease of disintegration by fouling [15]. The habitats and locations reported in this work were chosen since they represent well 797 the areas in which most plastic debris is found. The eulittoral test system is ideal to 798 investigate materials with a specific density below that of seawater that initially float at 799 the surface and can be washed ashore. This is the case for the most commonly used 800 conventional polymers polyethylene and polypropylene, and also all items with a 801 positive bulk buoyancy (e.g. bottles, foam), regardless of the material. The benthic 802 tests well represent the seafloor scenario, the sink for most plastic materials [33, 35] 803 due to biofilm formation. For this reason, the pelagic tests in which the materials are 804 artificially kept from further sinking despite biofilm formation could be considered the 805 least environmentally relevant. However, the tests provide important information about 806 the time during which the materials float in the water column. The test remains highly relevant to assess materials that are considered to be used directly in an aquatic 808 environment, for example in aquaculture systems. In this case materials should be 809 tested where they will be applied in the field. Technically, all test systems were proven to be stable and are suited for use over 817 several years to expose and test plastic films in coastal habitats. Some maintenance 818 to the systems such as regular cleaning is required, especially in areas of high fouling. 819 These test systems could also easily be transferred to other aquatic habitats and test 820 sites and be adapted to test objects instead of films. The test frames sufficiently 821 protected the samples and no mechanical damage of the specimens was detected and 822 we are convinced that under such test conditions the observed disintegration was 823 caused predominantly by biodegradation. 824 Mesocosms are well suited for manipulative experiments above lab flask scale. 825 Controlled parameters such as temperature and also nutrient content could be set to 826 optimum conditions to the physiological limit of the locations the matrices are collected 827 from. More replicates with a smaller size and the exclusion of light could lead to a more 828 consistent data set which is less affected by heterogeneity.       Anions, metals and metalloids, measured by analyitical lab