Propylene glycol – a useful capture preservative for spiders for DNA barcoding

The usefulness of propylene glycol as capture preservative in pitfall traps, with the aim of using the captured spiders for DNA barcoding, was tested. For this purpose a laboratory experiment on the conserving and/or denaturing effect of propylene glycol on mitochondrial DNA (COI) was set up. For the experiment 110 specimens of the common and abundant wolf spider species Pardosa lugubris were manually captured, killed and incubated from one to four weeks in either pure or watered propylene glycol or 70 % denatured ethanol. Rates of successful sequencing, following a standard protocol, did not differ between samples incubated in propylene glycol and in the more commonly used ethanol. Thus, within four weeks, propylene glycol did not significantly denaturize mitochondrial DNA. In two field studies, pitfall traps with propylene glycol captured more spiders than traps with acetic acid. The effect was significant only in one of two field trials, but then consistent at three different sites and the three dominant spider families. Based on these results and our operating experience, we recommend propylene glycol as a capture preservative for (pitfall) traps to obtain specimens for DNA barcoding identification.

country/region that was successfully barcoded (examples listed at https://www.bolgermany.de/newspublikationen,accessed January 2015; Miller et al. 2013).However, finding fresh specimens of the remaining species becomes increasingly difficult.This is especially true for those spider species not readily identifiable in the field, e.g.many linyphiids, but also for rarely found species (e.g.Clubiona kulczynski), for which too few data are available to find/resample them intentionally.DNA barcodes for some species which are rare only on a regional/national level, will certainly be added to a cross-national (international) DNA barcode library by other countries, but not including the target country would result in knowledge gaps regarding genetic variability.One promising way to complete the DNA barcode library, including regional representatives, is to target species in museum collections (Miller et al. 2013).Although successful standard sequencing decreases with the age of the specimens -i.e.time since collected, duration of preservation in 70 % ethanol) from 90 % (fresh) to 60 % (10-20 years preserved) to 12.5 % (more than 50 years) and also with the size of the specimenmuseum collections can contribute much to species completeness (Miller et al. 2013).
Notwithstanding all these efforts, DNA of some species and especially their genetic variability will only be provided by systematically collecting spider assemblages from all the different habitat types and bioregions.One important method to systematically sample spider assemblages are traps (e.g.pitfall traps, funnel traps/eclectors) and these usually use killing and preserving agents.Formerly widely used agents, like picric or acetic acid, formalin, saturated sodium chloride or specific mixtures like Galt's (5 % sodium chloride, 1 % saltpeter and 1 % chloral hydrate in water) unfortunately destroy/denature DNA (Vink et al. 2005, Stoeckle et al. 2010), whereas ethylene glycol (Gurdebeke & Maelfait 2002), propylene glycol (Vink et al. 2005), and also Renner solution (ethanol, glycerine, acetic acid, water) have shown to be suitable to preserve DNA (Stoeckle et al. 2010).Propylene glycol seemed to us the most promising substance due to its low toxicity to mammals (food additive) and its reasonable price (for the technical quality).We therefore tested propylene glycol (PG) for manageability in the field, capture and preserving ability, and lastly for its effect on DNA.
We wanted to test the usability of PG as capture agent in pitfall traps under simulated conditions (in the laboratory), using a large number of replicate samples (10 replicates per treatment, 30 replicates for PG versus ethanol, see Tab. 1), but also under real field conditions.After a non-systematic experiment using PG in pitfall traps and processing some of the captured spiders for barcode sequencing, we obtained an initial idea of success rates: for 74 % of specimens sampled in pitfall traps with propylene glycol (N = 367) COI sequences > 600 bp were obtained, in comparison to 92 % from recently hand collected and in ethanol (70 %) preserved material (N = 212).
Our initial hypothesis was that rain water would -at least in higher quantities -dilute the PG and would thus decrease DNA preservation.We wanted to check if the animals captured in PG during our normal field exposure times (two weeks) would still allow high success rates in Sanger-based DNA sequencing, whether small amounts of water entering the traps through rain or due to the hygroscopic nature of the preservative decrease the rate, and whether it is better to transfer the captured animals soon after collecting into ethanol instead of keeping them in PG until sorting and identification.Therefore we used incubation times of 1, 2 and 4 weeks for both preservatives.To imitate rain water intrusion in pitfall traps with PG as the capture liquid, we used treatment variants with 10 % and 50 % water; 10 % water was added in one variant from the beginning, in one variant 4 days after the spiders have been killed and conserved in pure PG.As we expected 50 % water to have a strongly negative effect, we did not test higher water contents.
Additionally, we wanted to check the capture efficiency of PG, i.e. whether PG had an (undesired) effect on the capture of spiders in pitfall traps, when compared to acetic acid, the preserving agent we had used in former studies based on pitfall traps.This was done in one field trial with 3 traps per agent in a relatively uniform habitat over 6 weeks and repeated in a second field trial with 3 traps per site and agent in three differing sites of a more heterogeneous habitat over 28 weeks.For both habitats we knew the spider fauna quite well from former studies.

Propylene glycol
From http://en.wikipedia.org/wiki/Propylene_glycol(accessed November 2014) the following information was extracted: Propylene glycol is a clear, colourless and hygroscopic liquid.It depresses the freezing point of water.In contrast to the closely related ethylene glycol -which is very toxic to humans and many animals -propylene glycol is considered safe (GRAS) by the U.S. Food and Drug Administration, and it is used as an humectant (E1520), solvent, and preservative in food and for tobacco products and in many pharmaceuticals, including oral, injectable and topical formulations.Propylene glycol is an approved additive for dog food and considered safe for dogs, but not for cats.LD50 is 20 mL/kg for most laboratory animals.PG is readily degradable, without special risks in terrestrial environments, but high levels of biochemical oxygen demand during degradation in surface waters can adversely affect aquatic life by consuming oxygen needed by aquatic organisms for survival.Surface tension is about half of that of water (35.6 mN/m at 20°C).More information can be found under http://www.propylene-glycol.com/(acessed November 2014).

Material and methods
We set up an experiment using manually collected specimens of Pardosa lugubris (Lycosidae) from one locality in Karlsruhe (20.6.2013),immediately transferring them into either propylene glycol, diluted propylene glycol, or 70 % denatured ethanol under the design summarized in Tab. 1.We used technical grade propylene glycol (C 3 H 8 O 2 ) from Herrlan PSM.All samples were always stored in a household refrigerator at the Staatliches Museum für Naturkunde Karlsruhe (SMNK) until finally cutting one leg of each spider and transferring it into absolute ethanol.These samples were then transported under cooled conditions to Stuttgart.DNA extraction was done in the laboratory of the Staatliches Museum für Naturkunde Stuttgart (SMNS) in March 2014 using the following procedure: Legs were used for parallelized, automated extraction in a Xiril Neon 100 robot, and amplification of the mitochondrial Cox1, 3' region (Simon et al. 1994) following the protocol developed by Ivanova et al. (2006).Unpurified PCR products were sent out to a sequencing facility (GATC GmbH, Konstanz).Contig assembly was handled using the software package Geneious (Biomatters, NZ) and lab data managed through a LIMS (Laboratory Information Management System).Specimens are deposited in the Arachnological collection of SMNK, DNA vouchers in the biobank of SMNS, field and laboratory data are linked through the database system Diversity Workbench (Triebel et al. 1999, Raub et al. 2012).
Pitfall traps were of the conventional type with plastic cups of 300 ml and an opening diameter of 67 mm, inserted in the ground, flush with the soil surface and filled with 100 ml preserving agent.Traps were protected against rain by non-transparent metallic roofs and to minimize capture of small vertebrates plastic funnels (opening diameter 1.6 cm, 30° angle) were inserted in the cups.Acetic acid was used in a 2.5 % solution with a drop of detergent added.Traps were emptied and refilled with fresh capture fluid every two weeks.
Statistics.Data from pitfall trapping were analyzed with a Manova (GLM) for the effects of fluid (capture agent) and site on the capture of individuals and species of all spiders, and the three dominant families with Statistica 9.0 (StatSoft Inc. 2009).

Results
All (100 %) 30 samples initially preserved in 70 % denatured ethanol and 93.75 % of 80 samples initially preserved in propylene glycol were successfully sequenced (i.e.COI sequence > 600 bp recovered).For only one of each treatment P4, PW2, PWW1 and two of PW1 or PCR was not successful, or the recovered sequence was < 300 (see Tab. 1).
Pitfall trapping in the meadow orchard during six weeks showed a positive, but not significant effect (F 1,34 = 0.08, p = 0.78) of PG (413 in PG versus 398 spiders in acetic acid) on the capture of spiders.

Discussion
The results may indicate a possible negative effect of water intrusion (PW-treatments) into propylene glycol on DNA preservation, albeit without statistical significance.None of the treatments destroyed mitochondrial DNA (COI) at a rate which renders PCR amplification unsuccessful in a standard procedure for barcoding.This was partly expected based on the published literature (Gurdebeke & Maelfait 2002, Stoeckle et al. 2010).However, before recommending the practical application of propylene glycol as a capture fluid in traps, several aspects had to be checked.Fortunately, addition of small to medium water quantities -which are likely to occur in the field due to the hygroscopic nature of PG, but also rain -did not lead to DNA degradation critical for DNA barcode sequencing.However, strong water intrusion or longer storage in watered PG should probably be avoided to guarantee high success rates.Using PG in pitfall traps protected against rain, leaving them two weeks in the field and then transferring spiders to 70-80 % ethanol for morphological identification seems feasible to maintain the potential for a later use of specimens for DNA barcoding.The very high success rates of both preserving agents in the laboratory trial in comparison with success rates from field samples shows that in the field a series of unknown and uncontrollable factors further influence the preservation of the specimens.The yield of field samples could probably be increased by early (i.e.before transfer to denatured ethanol) cutting of a leg and its preservation in 96 % non-denatured ethanol under stable, cool conditions.
The results are based on short exposure times: four weeks at most until transfer into the recommended preservation fluid (non-denatured 96 % ethanol).However our focus lay on testing the use of a capture preservative under field conditions and realizing the usual processing of faunistic/ecological samples in our lab.The use of PG as a long-time preservation fluid in collections, as an alternative to the very expensive non-denatured ethanol, should be investiga- ted over long-term experiments, including further aspects of long-term conservation.Vink et al. (2005) tested preservation in PG (and other preservatives) during six weeks and found PG to be superior to 95 % ethanol, at least for single copy genes.Regarding other aspects, the same authors reported a personal communication by M. J. Ramirez "It appears that propylene glycol may cause soft tissue shrinkage in specimens" (Vink et al. 2005).
In the light of the use of traps to capture spiders to complete the DNA barcode reference database of German species, we advocate the use of propylene glycol rather than acetic acid as a killing and preserving fluid in the field.Ethanol is not useful, due to its strong evaporation.The comparison of capture rates of PG with those of acetic acid shows that PG is more effective, perhaps due to its viscosity and lower surface tension compared to water.Although an attractiveness of particular capture liquids for individual species or groups has been shown and discussed by several authors (e.g.Adis & Kramer 1975, Adis 1979, Buchberger & Gerstmeier 1993, Gläser 2010), it was and is not feasible to compare even the more commonly used killing and preserving agents in their capture efficiency and undesirable effects in many habitats for the target taxa.However, we wanted to check at least how PG -as a promising agent under the new criteria of DNA preservation -behaves in this respect in comparison with the formerly used acetic acid.Most important here is the fact that PG was not selectively attractive to individual species or families and relative capture efficiency did not differ between the three sites.It can therefore be used to sample spider assemblages of different habitats, preserving the potential of identifying or verifying identifications (Vink et al. 2005, Blagoev et al. 2013) of trap sampled material through DNA barcoding, based on the comparison with a good reference database on German (European) spider species.

Fig. 1 :
Fig. 1: Effect of the preservative (capture fluid) on the capture of individuals and species in pitfall traps at three sites in montane heathland in the northern Black Forest.Results of a two-way Manova (GLM): effect of fluid (Wilks Lambda = 0.972, F(2, 233) = 3.34, p = 0.037) and site (Wilks Lambda = 0.93420, F(4, 466) = 4.03, p = 0.003) on the two-week capture of a. individuals per trap; b. species per trap; vertical bars show 0.95 confidence intervals.

Fig. 2 :
Fig. 2: Effect of the preservative (capture fluid) on the capture of the three dominant spider families in pitfall traps in montane heathland in the northern Black Forest.Results of a MANOVA (Wilks Lambda = 0.46, (F3, 14) = 5.34, p = 0.01); vertical bars show 0.95 confidence intervals.