Application of propylene glycol in DNA-based studies of invertebrates

High-throughput sequencing (HTS) studies on invertebrates commonly use ethanol as the main sample fixative (upon collection) and preservative (for storage and curation). However, alternative agents exists, which should not be automatically neglected when studies are newly designed. This review provides an overview of the application of propylene glycol (PG) in DNA-based studies of invertebrates, thus to stimulate an evidence-based discussion. The use of PG in DNA-based studies of invertebrates is still limited (n = 79), but a steady increase has been visible since 2011. Most studies used PG as a fixative for passive trapping (73%) and performed Sanger sequencing (66%; e.g. DNA barcoding). More recently, HTS setups joined the field (11%). Terrestrial Coleoptera (30%) and Diptera (20%) were the most studied groups. Very often, information on the grade of PG used (75%) or storage conditions (duration, temperature) were lacking. This rendered direct comparisons of study results difficult, and highlight the need for further systematic studies on these subjects. When compared to absolute ethanol, PG can be more widely and cheaply acquired (e.g. as an antifreeze, 13% of studies). It also enables longer trapping intervals, being especially relevant at remote or hard-to-reach places. Shipping of PG-conserved samples is regarded as risk-free and is authorised, pinpointing its potential for larger trapping programs or citizen science projects. Its property to retain flexibility of morphological characters as well as to lead to a reduced shrinkage effect was especially appraised by integrative study designs. Finally, the so far limited application of PG in the context of HTS showed promising results for short read amplicon sequencing and reduced representation methods. Knowledge of the influence of PG fixation and storage for long(er) read HTS setups is currently unavailable. Given our review results and taking difficulties of direct methodological comparisons into account, future DNA-based studies of invertebrates should on a case-by-case basis critically scrutinise if the application of PG in their anticipated study design can be of benefit.


Introduction
DNA-based high-throughput sequencing (HTS) approaches such as DNA metabarcoding have lately revolutionised our ability to comparatively assess and monitor biodiversity over large geographical scales and at an unprecedented rate (Taberlet et al. 2012;Aylagas et al. 2016;Leese et al. 2018;Compson et al. 2020). Of particular relevance in practice are declining costs per sample, which are driven, among other things, by laboratory au-tomatisation, comprehensive parallelisation of samples (i.e. multiplexing) and ever decreasing sequencing expenses (Leese et al. 2018). On the other hand, large-scale environmental programmes require a high number of appropriately conserved samples. Assuming that the number of samples to be processed within a scientific study or environmental program will continue to increase, also field costs per sample will account for an increasingly large financial share. It is therefore important to develop strategies to counterbalance the increasing costs of field work, e.g. by reducing the time spent in the field, lowering hands-on times per sample by automatisation or adjusting collection and storage conditions (Holderegger et al. 2019;Ssymank et al. 2018).
The fixative (or trapping / killing agent) is a crucial parameter for DNA-based studies as it has to secure the integrity of the DNA from the very beginning (see e.g. Stein et al. 2013). Simultaneously, its costs are directly linked to the number of samples to be processed and to the volumes used. It is not unusual for larger studies to require several hundred to thousand litres of fixative.  identified lab grade ethanol (>95%) to be the fixative of choice in the majority of DNA metabarcoding studies conducted between 2015 and 2019. Yet, lab grade ethanol is very expensive, has a high volatility, is not always easy to obtain and due to its flammability can cause problems during sample transportation and shipping. Alternative fixatives exist but are not widely applied, e.g. due to misunderstandings of chemical properties, tradition or simply lack of knowledge and availability (Weeks Jr and McIntyre 1997;Thomas 2008;Nagy 2010;Stoeckle et al. 2010;Gossner et al. 2016;Mahon et al. 2017).
As such an alternative, propylene glycol (or propane-1,2-diol; here further abbreviated as PG) has a number of characteristics that potentially render it beneficial in large-scale DNA-based HTS bioassessment and biomonitoring programmes: (i) low acquisition cost but high general availability, as it can be bought, for example, as low-budget antifreeze in specialist car dealers or as an additive from the cosmetic or food industry (additive E1520), (ii) non-toxicity (i.e. considered as a GRAS (generally regarded as safe) material), (iii) very low volatility, (iv) environmental safety, (v) risk-free transport of samples according to the regulations of the International Air Transport Association (IATA), and (vi) ability to ensure DNA integrity as well as to preserve most morphological characteristics (Thomas 2008;Nagy 2010).
Propylene glycol is a well-established agent in molecular cryobiology, notably used for the cryopreservation of sperms and cell cultures (Bank and Brockbank 1987;Hezavehei et al. 2018). For the study of invertebrates, it has been proposed as the chemical of choice in a standardised pitfall trap design for monitoring ground-active arthropod biodiversity (Brown and Matthews 2016;Hohbein and Conway 2018) and is applied as such within the North American National Ecological Observatory Network (NEON; Gibson et al. 2012;Hoekman et al. 2017) and the carabidologist community (Kotze et al. 2011). PGbased pitfall traps were also proposed as a standard and minimally disturbing method to investigate the subterranean fauna of the mesovoid shallow substratum (López and Oromí 2010; emptied every six months). Since 2001, PG fixation is applied by the Soybean Aphid Suction Trap Network (STN; Lagos-Kutz et al. 2020) for monitoring the 'aerobiological soup'. In the context of environmental genomic studies, Matos-Maraví et al. (2019) proposed PG as the fixative of choice for mass samples originating from pitfall traps and flight interceptions, but neither tests nor metadata had been published along with this statement.
The aim of this review is to summarise the findings of studies investigating the application of PG for DNAbased analyses of invertebrates. We will distinguish its application as a fixative (during sample collection) and as a preservative (for sample storage and curation). The collated information should help to transfer the available knowledge to the wider community, stimulating an evidence-based discussion on how to further reduce costs for DNA-conform sample collection and curation in larger environmental programmes by exploring alternative fixation and preservation agents.

Literature search
A topical core literature research was conducted on 07.10.2020 within the ISI Web of Science (WoS) and screening 'all databases'. The following search strings were investigated: (1) "propylene glycol" AND "DNA" AND "invertebrate*", (2) "propylene glycol" AND "DNA" AND "insect", (3) "propane-1,2-diol" AND "DNA" AND "invertebrate*" and (4) "propane-1,2-diol" AND "DNA" AND "insect". On the very same day, and because the WoS search only will detect literature records which are ISI-listed and only will retrieve hits in case the search string terms appear in the title, abstract, topic or as keywords, a complementing Google Scholar (GS) search was performed (as e.g. proposed by Piasecki et al. 2018). The GS search enables in-text searches and will also retrieve grey literature records such as field, laboratory and shipping protocols, pre-prints, theses and agency reports. The same four search strings were entered, selecting the 'exclude patents' option in GS. The keyword 'sperm*' was excluded in all searches as PG is frequently applied during the cryopreservation of sperms which would have resulted in a very high number of irrelevant hits. The key word 'invertebrate*' was excluded in the search strings (2) and (4) to not again screen insect literature records retrieved from (1) and (3).
The four search strings yielded the following number of literature records within the ISI WoS: (1) 20, (2) 8, (3) 0 and (4) 1. From those, only records which transparently stated the application of PG during specimen handling have been considered, resulting in a total of only 12 studies. Google Scholar search strings yielded the following constant number of literature records, regardless of whether the search was performed via two different IP-addresses in two countries (Luxembourg and Germany), within two different Google profiles, using a private browser window or being logged on/off from personal Google profiles: (1) 509, (2) 1930, (3) 23 and (4) 432.

Results
The majority of studies had to be excluded, a) because they used PG as a trapping material for invertebrates but only Figure 1. Cumulated number of studies which have used propylene glycol either as a fixative or preservative in the DNAbased analysis of invertebrates. Arranged by the taxonomic groups in focus. In cases where several of the indicated groups were targeted, the study was placed into the category "others". N = 79, as of 07.10.2020. cited literature referring to the application of DNA-based tools, or b) as they referred to yeast species molecularly analysed. Furthermore, reviews or studies which only dealt with the DNA analysis of e.g. parasites or gut contents of invertebrates were omitted, but will be discussed. Doctoral and master theses which were subsequently published as scientific articles were counted as a single entry.
A total of 79 publications was retrieved (Table 1, Fig. 1). Although the restricted number of literature records obtained exemplifies that the number of studies using PG as a fixative or preservative for subsequent molecular DNAbased analysis of invertebrates is not widely established (but an increase can be noted), still some patterns can be inferred. The analysed studies almost exclusively investigated terrestrial taxa, with Robinson et al. (2020; freshwater bulk sample) and Cordero et al. (2017;aquatic insects) being notable exceptions. The first studies using PG for DNA-based analysis of invertebrates originate from 2003 and 2005. Two of them focussed on honey bees (Rubink et al. 2003;Coulson et al. 2005) while Carter (2003 performed a comprehensive study on 'The effects of preservation and conservation treatments on the DNA of museum invertebrate fluid preserved collections' using a terrestrial isopod species as reference. Those studies were accompanied by early publications on arachnids (Vink et al. 2005) and hemipterans (Scott et al. 2007). From today's perspective, studies focussing on Coleoptera (n = 24 studies, 30%) and Diptera (20%) dominated the dataset (Fig. 1). Although Diptera was the second largest of the individual groups retrieved, the first study was only published in 2012 and most of the records originate from the same team of authors.

Application of propylene glycol as a fixative
Our results indicate that PG is widely applied as a fixative in a variety of passive trapping methods (e.g. pan trap, funnel trap, aerial pitfall trap; baited and unbaited) and for various organism groups (mainly beetles and flies, but also spiders, bees and aphids). Likewise, actively collected single specimens or -in a few cases -invertebrate bulk samples were fixed with PG (e.g. Bowser et al. 2017Bowser et al. , 2019Cordero et al. 2017;Jusino et al. 2019;Robinson et al. 2020;). This taxonomically widespread application mimics the use of PG in traditional non-DNA-based research and application domains (primarily pitfall trapping; see Hohbein and Conway 2018) now entering the molecular field. Research communities around aphids, spiders, carabids and tephritid fruit flies contributed most of the studies.
Amongst others, passive trapping intervals are determined by the accessibility of the sampling location, the volume of the trapping containers, the evaporation rate of the fixative and local environmental parameters such as humidity, temperature, rainfall or UV-exposure. In case of excessive heat and high temperatures (arid and hot), traps with ethanol or (salted) water frequently dry out and either have to be visited and re-filled more frequently, or short(er) trapping periods have to be chosen. The evaporation rate of PG is >500-times lower than for ethanol (Moreau et al. 2013). As such a low volatile agent, PG-equipped traps can retain their volumes more or less constant over several weeks to months (Hohbein and Conway 2018). However, long-term UV exposition may decompose PG into water, acetone and 2-propanol (i.e. isopropanol) (Nakahama et al. 2019). In interaction with its hygroscopic nature, even apparently large fixative volumes at the end of a long trapping period under summer conditions might contain a Table 1. Overview of studies which have used propylene glycol either as a fixative or preservative for the DNA-based analysis of invertebrates. RT = room temperature; PG = propylene glycol; n.a. = not applicable or not available. Studies found by both the ISI Web of Science and Google Scholar searches, are marked in bold. All others were only detected by the latter search engine. N = 79, as of 07.10.2020.

Study
Year Taxon Carter (2003)  all treatments produced a large number of high-quality loci (>4,000, ~20×). In comparison, the two PG-preserved species showed average DNA concentrations, but higher than average mean locus depths and lower than average mean numbers of polymorphic loci 3 days (RT) evaporation step prior to DNA extraction was ommitted as PG does not inhibit PCR; communities are highly similar and even showed higher proportion of arthropod reads and higher richness than samples conserved with >99% lab grade ethanol not to be underestimated amount of water and other compounds (Nagy 2010;Goetze and Jungbluth 2013;Borges et al. 2016), but chemical interactions and their effects on medium-to long-term DNA integrity are largely unknown. In addition to high temperatures, water dilution of PG in principle leads to a generally lower DNA integrity (Stevens et al. 2011;Höfer et al. 2015;Patrick et al. 2016;Nakamura et al. 2020), since water can lead to the hydrolysis of nucleic acids. Since most reviewed studies did not indicate the nature of the propylene glycol used, or field and storage temperatures, it was difficult to compare study outcomes. Nevertheless, in several studies, PG was used in the range of 20-80% as a trapping agent and traps deployed for a duration of 1-2 weeks. Specimens collected and fixed under these conditions were still sufficiently preserved for DNA analysis (Coulson et al. 2005;Sonoda et al. 2010;Hoekman et al. 2017;Angelella et al. 2019). In one of the most extreme cases, samples were fixed for 1 month in pitfall traps containing PG-antifreeze. The traps further experienced heavy rainfall and sediment wash-in. Nevertheless, Ramírez et al. (2019) successfully amplified six marker genes for the spider family under investigation and were able to morphologically investigate the specimens at hand. Propylene glycol is in accordance with the Dangerous Goods Regulations of The International Air Transport Association (IATA). This means that PG-fixed samples are suitable for direct shipping and do not have to be transferred to another chemical agent on the spot. This characteristic was especially important for studies in remote areas (e.g. Schutze et al. 2012;Haase and Zielske 2015;Patrick et al. 2016;Bagnall 2016;Boontop et al. 2017a). In addition, it was seen as advantageous that PG can be relatively easily and cheaply obtained as an antifreeze in many parts of the world -a fact that does not always apply to absolute ethanol. As a major drawback, however, PG-based antifreezes might be regarded as special waste. As such, it can be prohibited to introduce them into local septic systems and the natural ground or to dispose them as domestic waste (Thomas 2008;Renaud 2012). Some national programmes might even prohibit the general use of antifreezes, as they can come along with additives such as lubricants, buffers, corrosion inhibitors and anti-foaming agents, whose impacts on natural environments are often not totally understood -or are considered carcinogenic (e.g. Tolytriazole, see Thomas 2008). Yet, much of the concern and regional bans relate to ethylene glycol-based antifreezes, which are toxic to humans and impose environmental risks. Readily available food-grade PG-based formulations such as from swimming pool or recreational vehicle antifreezes are generally regarded as safe (GRAS) material. They metabolize to lactic acid or substances of the Krebs cycle, which are natural metabolic products in the environment (Thomas 2008;Skvarla et al. 2014). Nevertheless, the waste-disposal issue imposed by using antifreeze fixatives (as it is true for larger quantities of ethanol as well) should be understood beforehand and appropriate measures taken.

PG specificities Fixation step Preservation step Central outcomes Condition(s) Duration Condition(s) Duration
Another relevant aspect refers to a potential catch bias caused by the fixative. Although Schmidt et al. (2006) have not integrated PG in their test, they showed that capture efficiencies of commonly used fixatives in pitfall trapping of spiders and carabids can greatly vary. Adding to this, Weeks Jr & McIntyre (1997), Calixto et al. (2007) and McCravy and Willand (2007) demonstrated that PG might not only affect the size of the sample but also its taxonomic composition. Höfer et al. (2015) reported that pitfall traps deployed with PG captured significantly more spider species, but were not selectively attractive for par-ticular taxa. Nor was the pattern consistent across all sampling sites. Reliable reference data for PG capture rates still seem to be too small to make valid statements, but the community should be aware that such a taxonomic bias could exist. Additionally, mammals can be attracted by the use of PG, disturbing pitfall traps and biasing sampling designs (e.g. Aristophanous 2010 and own observations).

Application of propylene glycol as a preservative
Propylene glycol was only occasionally used as a medium-term preservative or storage medium. After PG fixation, most samples were stored in ≥95% ethanol until DNA extraction or PG-fixed samples directly analysed within a few days when retrieved from the field. Nevertheless, Nakamura et al. (2020) highlighted the potential use of PG as a chemical agent that can be applied from trapping to storage and for various taxonomic groups, so that hands-on times can be shortened as well as labour and equipment costs reduced (i.e. no specimen picking, circumventing transfer into another solution and container). The authors compared the COI-amplification rates of dipterans, hymenopterans and coleopterans preserved in 98% PG and 99.5% ethanol over a period of 2 weeks to >6 months at room temperature, concluding that DNA might be more long-term stable in PG than in ethanol. On the contrary, Patrick et al. (2016) tested various storage agents (including 99.5% PG and 97-100% ethanol) applicable in remote areas on three dipteran species under different temperatures. Although the experiment ended after 15 days, their results suggest that keeping PG fixed samples as cold as possible (e.g. packed in ice-filled boxes from hotel bars) is important to ensure short-to mid-term DNA integrity.
If we adopt the results to common practices of sample storage, it tells us that PG-preserved samples should preferably be stored cool and dark just like ethanol-preserved samples. Yet, DNA quality and quantity of long-term stored, chilled PG-preserved samples should be investigated in further detail.

Application of propylene glycol in HTS studies
Sufficiently high DNA quantities and DNA qualities are prerequisites in HTS studies. Lienhard and Schäffer (2019) evaluated DNA quality and quantity of ethanoland PG-preserved oribatid mite species (<1 mm, preserved for several weeks), originating from seven DNA isolation methods suitable for high-throughput DNA sequencing. Although some study parameters had a significant effect on DNA quantity and quality, results for specimens preserved in PG or absolute ethanol generally suggest a high comparability. Similarly, Carter (2003) investigated the molecular weight spectra of double-stranded DNA (dsDNA) for specimens of the Rough Woodlouse Porcellio scaber preserved in PG, ethanol, ethyl acetate and 2-ethoxy ethanol for 12 months at room temperature. Ethanol-and cryo-preserved specimens provided the best quality and highest concentration of high molecular weight dsDNA for the investigated time period (with a remarkable drop after 24 months for ethanol). Yet, DNA quality (in terms of degradation) and DNA quantity (in terms of concentration) of specimens stored in PG were decreased, especially for longer fragments (>5 kbp).
However, specimens were stored at room temperature, which showed a strong degradation effect in Patrick et al. (2016) compared to samples which were kept cool.
Systematic studies analysing the impact of PG fixation and preservation in the context of HTS are widely lacking. Still, first case studies indicate a high applicability of PG fixation for short read amplicon sequencing (metabarcoding; Bowser et al. 2017Bowser et al. , 2019Lefort et al. 2017;Jusino et al. 2019;Robinson et al. 2020). In particular Robinson et al. (2020) tested the effect of PG antifreeze fixation of homogenised mock communities and benthic bulk samples. Their COI metabarcoding results indicate a generally high comparability of communities from ethanol and PG fixed samples. The latter even produced a higher proportion of arthropod reads and a higher richness of exact sequence variants (ESVs) when compared to ethanol samples.
Besides studies on short read amplicon sequencing, the applicability of PG was shown for short read reduced representation methods (RAD-sequencing; Perry et al. 2017;Angelella et al. 2019;Ballare et al. 2020). In particular, Ballare et al. (2020) conducted a comparative setting testing the impact of different field sample methods for SNP detection in wild bees, including PG-filled Vane traps. While a suitable DNA concentration and high locus depth were found for PG-fixed specimens, the number of loci recovered was comparatively low. Still, the two targeted wild bee species on average possessed more than 10,000 loci with a mean locus depth of >70. One potential cause for the lower than average mean number of polymorphic loci might be an incomplete PG evaporation, leading to an insufficient ethanol preservation prior to pinning. Alternatively, PG-stored samples were kept at room temperature, whereas other samples which showed higher numbers of polymorphic loci were stored at -20 °C.
To the best of our knowledge, no study used PG-fixed or -preserved specimens for invertebrate genome sequencing. Our assumption, however, would be that if samples are fixed and stored under optimal conditions, genome sequencing based on short read lengths should be possible. However, how ultra-long sequencing (e.g. Nanopore) will be affected by potential DNA degradation effects of PG remains unclear and should be explicitly addressed.

Integrative use cases of propylene glycol-conserved invertebrate samples
Besides the possibility to perform DNA-based analyses (e.g. microsatellite fingerprinting, DNA barcoding, metabarcoding and RAD-seq) directly on the invertebrates trapped or stored in PG, the samples seem suitable for a variety of research designs. Firstly, it has to be highlighted that PG-conserved specimens demonstrate a reduced shrinkage effect and specimens often remain appropriately conserved for morphological examina-tions (Thomas 2008). More so, PG can be even added to ethanol-preserved specimens to retain flexibility and to reduce shrinkage effects without compromising DNA integrity and storage conditions (Boaze and Waller 1992;Carter 2003), e.g. for improving morphological determinability and dissection conditions (Karanovic et al. 2012(Karanovic et al. , 2016Martin 2016;Perina et al. 2018;Herrera Russert 2019). This 'relaxation' property of PG was exploited by several studies examining DNA markers and morphology of the very same specimens (Villacorta et al. 2008;Krosch et al. 2013;Boontop et al. 2017b;Gregoire Taillefer and Wheeler 2018;Grando et al. 2018). Furthermore, PG-conserved invertebrate samples were highly suitable for related integrative study designs, which are becoming increasingly popular in modern biodiversity studies. As such, Hu et al. (2017) investigated insect-associated microbiota by 16S rRNA amplicon sequencing of ants, Tremblay et al. (2018Tremblay et al. ( , 2019 screened phytopathogenic fungal propagules captured in insect traps via metagenomics, and Lynggaard et al. (2019) analysed bulk arthropod Malaise trap samples to detect vertebrates via 12S and 16S DNA metabarcoding. Molecular gut content analyses of PG-fixed invertebrate samples were e.g. performed for mosquitos (Muturi et al. 2018), fruit flies (Diepenbrock et al. 2018), carrion-feeding flies (Bagnall 2016) and diverse predatory groups (Murtiningsih 2014;Mabin et al. 2020). Endoparasites were investigated by e.g. Sokolova et al. (2010;microsporidians  Finally, an addition of PG to pheromone/food traps might extend the lifetime of kairomones (Faleiro et al. 2016). Propylene glycol fixation of samples thus seems to be promising for a variety of integrative study designs.

Conclusions
There is currently only limited scientific literature on the use of PG for DNA-based analyses of invertebrates available, and even less so in the context of HTS. However, the investigated studies indicate that PG can be a versatile and worthwhile alternative for sample fixation (and potentially preservation) of various organism groups and in a range of methodological setups. Yet, generally valid statements about fixatives and preservatives are difficult to make (Nagy 2010;Short et al. 2018), and can be biased by comparing agents with contrasting water conditions affecting DNA integrity (see Nakamura et al. 2020), varying field vs. laboratory humidity conditions (due to the hygroscopic nature of PG) and storage temperatures or by confusing chemicals (e.g. Vaudo et al. 2018).
Future studies which plan the application of PG should critically scrutinize their trapping, specimen and storage conditions. For how many days are traps deployed? How will humidity, precipitation, UV exposure and temperature conditions in the sampling area affect the fixative? Do the targeted organisms allow for an easy tissue penetration by the fixative (e.g. soft-bodied vs. sclerotinised specimens)? Can PG reduce hands-on times and yet overall costs (e.g. no sample transfer for shipping; no evaporation prior to DNA isolation) (see e.g. Robinson et al. 2020)? In many cases, and in particular for new largescale DNA-based monitoring programmes, one should not simply go by tradition but perform environment-and target group-specific tests and cost calculations before deciding upon the most suitable fixative.