Adhesive Tape Strips and PVA–Borax Hydrogels as Alternative Cleaning Methods to Remove Fungal Proliferation on Canvas Support of Paintings

TL;DR

This study explores new methods using adhesive tape and hydrogels to clean fungi from paintings, showing promising results compared to traditional techniques.

Contribution

The paper introduces and evaluates PVA-B gel and adhesive tape strips as novel, effective methods for fungal removal from canvas artworks.

Findings

PVA-B gel effectively removed fungi from canvas surfaces and subsurfaces without damaging fibers.

PVA-B gel with Cyrene™ showed antifungal activity comparable to that with clove essential oil.

Adhesive tape strips slightly outperformed micro-aspiration in fungal removal for most cases.

Abstract

Two commercial adhesive tape strips (Fungi-TapeTM and Filmoplast® P) and a polyvinyl alcohol–borax (PVA-B) gel were tested as novel physical cleaning alternatives to micro-aspiration for removing visible fungal colonisation from a cotton canvas. In addition, clove essential oil (CEO) and Cyrene™ were incorporated in the PVA-B gel for testing the potential of each to improve fungal cleaning. For the trials, canvas mock-ups were separately inoculated with two fungal species identified as Penicillium chrysogenum and Aspergillus westerdijkiae. Removal of fungi and related impacts were evaluated by DOM, FESEM, ATR-FTIR and ImageJ software. Inhibition of fungal spores and residual growth were assessed by in vitro growth tests and CLSM. Removal of A. westerdijkiae was more effective than removal of P. chrysogenum, especially for dense coverage. Both tape strips removed slightly more fungus than micro-aspiration, except for dense coverage of P. chrysogenum. The PVA-B gel, both with and without CEO or Cyrene™, yielded the best (similar) results, removing the fungal material found on the surface and subsurface of canvas without damaging the canvas fibres. Although further testing is required, the antifungal activity of PVA-B gel+ Cyrene™ seems comparable to that of PVA-B gel+ CEO, the former being especially effective against A. westerdijkiae.

Figures

Full text

1. Introduction

Fungal colonisation compromises canvas integrity, causing physicochemical, mechanical and aesthetic alterations that hamper long-term conservation efforts [1]. Xerophilic or xerotolerant fungi such as Aspergillus, Cladosporium and Penicillium represent the main risk to canvas conservation [2,3]. These fungi have a low requirement for water and can grow at low water activity (a_w_, ranging from 0 to 1). Fungi cannot generally develop at a_w_ below 0.7 [4], except for xerophilic fungi, which grow at a_w_ values between 0.60 and 0.80 [5]. In addition to the water activity, temperature and pH are also associated with the potential risk of spore germination and/or mycelial growth. Bastholm et al. [6] hypothesised that low indoor temperatures in museum repositories may cause locally elevated a_w_ suitable for fungal germination, even at RH < 60%. However, despite the increased interest in the control and prevention of biodeterioration of canvas paintings over the last decade [7,8], few studies have specifically addressed the removal of fungi from canvas. Fungal colonisation can be found on the surface, subsurface and within canvas fabric. Fungi are composed of networks of hyphae, called mycelia, which form a weft or tissue and vary according to function. The vegetative mycelium adheres to the substrate on which the fungus is growing, absorbing nutrients and water for growth. The aerial or reproductive mycelium grows above the substrate and is important for reproduction, carrying conidia and fungal spores. The vegetative mycelium can be introduced, to a greater or lesser extent, in the cotton canvas fabric, and elimination of the fungus is therefore difficult. Xerophilic fungi physically anchor to the cotton canvas by using turgor pressure to wedge hyphae into cellulose pores and thigmotropic growth to weave a dense, interlocking mycelial network around individual cotton-fibre groups [2,3]. Chemically, they anchor to cellulose fibres by secreting an extracellular matrix products and surface-active proteins, called hydrophobins, which enable hydrogen bonding and hydrophobic interactions between the fungal chitin–glucan complex and cellulose fibres’ hydroxyl groups [2,3]. So, material-based physical cleaning methods must manage to insert as much material as possible into the fibres and pores to remove as much of the fungus as possible, without damaging the canvas fibres.

It is impossible to design a universal cleaning method, due to the variety of canvas materials and differences in twist and crimp dimensions [9], the number of combinations in which they are used to produce historically and artistically significant artefacts, and the diversity of conservation states and tools available to restorers. In addition, some methods of removing fungal colonisation may not be applicable in more specific contexts. Therefore, it is important for restorers to have available multiple methods and protocols, to enable them to customise the removal of biodeteriogenic fungal strains from cotton canvas. This is the context within which the present work is framed.

Micro-aspiration is traditionally used for the mechanical removal of fungi from canvas because of its simplicity and the immediacy of the results [10]. It is not used on artefacts affected by active colonisation. Hydrated and active vegetative or reproductive mycelia are, in fact, more difficult to remove and can rapidly cause secondary outbreaks. The mycelia must be inactivated by gradual changes in the environmental conditions (e.g., RH reduction) under which the artefact is conserved. Altering these conditions reduces the bioreceptivity (ability of a material to be colonised by living organisms, [11]) while gradually rendering the colonising fungi inactive. If the power and/or duration of micro-aspiration are not appropriate for the intervention, using this method can lead to deformation of the fabric, dispersion of fungal colonisation and inadequate removal of fungi. Protocols that overcome these drawbacks involve the use of high-efficiency particulate air (HEPA) filters in the micro-aspiration system and gentle mechanical brushing with a soft brush [10].

In the search for alternative methods to micro-aspiration to treat fungal colonisation of cotton canvas, two commercial adhesive tape strips (Fungi-Tape^TM^ and Filmoplast^®^ P) and a polyvinyl alcohol–borax (PVA-B) gel were tested using the ‘press and peel’ cleaning procedure (pressing the tape or gel onto the contaminated surface and then peeling it off along with the unwanted material) as the removal method. In addition, clove essential oil (CEO) and Cyrene™ were separately added to the PVA-B gel to potentially enhance the performance of this treatment.

Adhesive tape strips (amongst which Scotch tape is the best known) have been commercially available since the late 1920s and used (mainly since the 1970s) by conservators and heritage scientists as a rapid and convenient tool, in four primary tasks: (1) to repair and protect paper artwork [12]; (2) to characterise conditions and assess weathering and also to assess the performance of surface coatings and surface consolidation treatments, mainly for stone substrates and related materials such as mortars and plasters [13]; (3) to sample algae, cyanobacteria, fungi, bacteria and diatoms from the surfaces of different solid substrates, including plastic, metal, brick, mortar, various types of rock and paintings [14]; and (4) to remove unwanted material from surfaces; however, as far as we are aware, no studies of this particular topic have yet been published. Gel-based cleaning methods for heritage have been increasingly adopted, especially in the last decade, due mainly to their ability to remove easily by mechanical peeling the unwanted materials and to confine liquids (such as solvents and biocides) within the gel matrix, limiting internal penetrations [15,16]. Conventional physical polymer-based cleaning gels, such as Carbopol^®^, Permulen^®^, Ethomeen^©^, Gellan and Klucel^®^ G, can, however, be difficult to fully remove after treatment, particularly when applied to porous substrates. They left residual gel materials, which often requires the use of pure organic solvents, representing a well-recognised and controversial limitation [17,18]. Agar (natural polysaccharide) gels have been used frequently in recent years precisely because, among other advantages, they leave less waste behind [19,20]. Polyvinyl alcohol (PVA) hydrogels have gained interest in the conservation and restoration field because of their low toxicity, high biocompatibility, biodegradability and combined chemical and mechanical cleaning properties (see, e.g., [21] and references therein). Polyvinyl alcohol–borax (PVA-B) gels have previously been used for cleaning purposes, to remove aged coatings [22] and metal corrosion products [23], because their high elasticity and malleability enable safe removal through exfoliation action that leaves very little residue on the surfaces to which they are applied [24]. Controlled cleaning action is also guaranteed and is more suitable for cleaning porous materials if applied for an adequate contact time, which is often difficult to achieve with rigid gels. In addition, PVA-B gels have a three-dimensional structure with a space that allows the inclusion of other substances’, such as organic solvents, biocides and surfactants [25,26], loading capacity depending on the hydrolysis degree of the PVA polymer [24,25,26]. Regarding the rheological properties, PVA–borax gels exhibit a storage modulus typically between 500 and 600 Pa, with a higher storage modulus (G′) compared to the loss modulus (G″). These gels also show shear-thinning behaviour and have a primary water content of 95–96%, with the capacity to load up, for instance, to 17.5% xylene and up to 25% non-solvent fluids before significant liquid leakage occurs [27,28,29].

In easel-painting restoration, PVA-B gels have mainly been employed for the cleaning of pictorial layers, particularly for aged or oxidised varnishes removal [27,28]. Lazidou et al. [27] evaluated solvent-free PVA-B gel systems for the removal of natural resin varnishes, concluding that effective varnish removal could be achieved using a contact time of 30 s. Carreti et al. [28] evaluated PVA-B gels incorporating 15% of acetone to assess their cleaning efficacy on a highly aged shellac varnish present on a 15th century tempera painting. The authors concluded that 5 min of contact between the PVA-B gel with acetone and the painted surface was effective in achieving varnish removal. The use of PVA-B gels in canvas or textile conservation has not yet been extensively investigated. Despite this, a recent study of Al-Emam et al. [16] explored the application of PVA-B gel as cleaning system for soot-contaminated silk textiles with different formulations, including water-based systems as well as gels loaded with ethanol or a non-ionic surfactant. The authors concluded that poly(vinyl alcohol)–borax–agarose double network gels represent an effective cleaning system for soot-contaminated silk.

Beyond their use in cleaning of abiotic components, PVA-B gels have increasingly been explored for applications related to biodeterioration. Boccalon et al. [30] achieved good results using two applications of PVA-B gels containing silver nanoparticles (AgNPs), silver nanoparticles–silver chloride (Ag/AgCl) or thyme essential oil (TEO) to treat stone heritage colonised by bacteria, fungi and algae. As a result of recent research, Estrela Monreal et al. [31] developed several formulations incorporating up to 70% ethanol into the PVA-B matrix, with successful results regarding the removal of Aspergillus sp. and Penicillium sp. from gesso-primed plywood panels with graphite line drawings. Similarly, Sala-Luis et al. [32] obtained good results by applying PVA-B loaded with a BFA (bactericidal, fungicidal and algicidal) biocide, containing benzalkonium chloride, benzisothiazolinone and methylisothiazolinone, to gypsum sculptures affected by Penicillium spp. fungi. Alternative (non-quaternary ammonium compounds form the basis of most chemical biocides) potential antifungal agents of natural origin to potentially enhance fungal cleaning, such as CEO and Cyrene™, can also be incorporated in the matrix. CEO, the main component of which is eugenol, has been found to be a potential green cleaning agent [33], successfully tested in vitro against isolated fungi from canvas paintings [34]. Cyrene™ is similar to dimethyl sulfoxide (DMSO) solvent in polarity and solubilizing power but has a green profile, proving to be a relatively safe bio-based substitute in several cleaning studies [35]. DMSO-based gel has been considered a more practical alternative to biocide treatments against meristematic black fungi in restoration and conservation practices [36]. Celi et al. [37] also indicate that essential oils (EOs) and DMSO-based gels, according to the scarce information about their use against fungi, seem to be as effective as the traditional biocides.

This study aimed to evaluate three potential mechanical methods, one also enhanced with antifungal agents, as alternatives to micro-aspiration, for treating fungal growth on cotton canvas. Two commercial adhesive tape strips (Fungi-Tape^TM^ and Filmoplast^®^ P, not previously used for cleaning with published studies) and a polyvinyl alcohol–borax (PVA-B) gel, alone and combined with clove essential oil (CEO) or Cyrene^TM^ (neither of which have previously been tested for fungal removal or antifungal activity on canvas) were tested using the ‘press and peel’ cleaning procedure to remove two fungal species (Penicillium and Aspergillus) inoculated separately on cotton canvas to replicate contamination. Both fungi (identified in this study) were previously isolated from the painting on the ‘Santa Rita gravemente enferma’ (Gravely ill Saint Rita) canvas (Figure S1) held in the Church of Los Santos Juanes in Valencia, Spain.

2. Results and Discussion

The results and discussion of the study are structured to first show the characterisation of fungal strains and materials, the effectiveness of the treatments, and finally the possible side effects. ATR-FTIR results, together with gas chromatography–mass spectrometry (GC-MS) in the case of CEO and Cyrene^TM^, were used to analyse the materials. Fungal removal and related impacts were assessed by digital optical microscopy (DOM), field emission scanning electron microscopy (FESEM) and attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy. Likewise, the efficacy of fungal removal was estimated using ImageJ software (version 1.54e). Growth inhibition and antifungal activity, respectively, were assessed by in vitro growth tests and by live/dead cell staining and confocal laser scanning microscopy (CLSM). Finally, the potential benefits and drawbacks of the six treatments tested are summarised in a table: micro-aspiration, Fungi-Tape^TM^, Filmoplast^®^ P, PVA-B gel, PVA-B gel+ CEO and PVA-B gel+ Cyrene^TM^.

Identification of the two fungal strains used in the study revealed the putative Aspergillus sp. (isolated sequence AN1) to be Aspergillus westerdijkiae and the putative Penicillium sp. (isolated sequence PR1) to be Penicillium chrysogenum (Figure S2). Aspergillus westerdijkiae, formerly assigned to A. ochraceus [38], is a filamentous fungus mainly known to cause food spoilage by colonising agricultural crops and food products, although it can also be present in household dust and indoor air in residences and workplaces [39]. It can proliferate over a wide temperature range (15–30 °C), although growth is optimal at 27 °C and water activity (a_w_) between 0.93 and 0.97 [40]. Strains of A. westerdijkiae have been found to be able to grow at a_w_ < 0.89 on coffee-cased medium and in stored coffee [41] and also to undergo slow growth at a_w_ values between 0.85 and 0.90 [42]. According to Han et al. [43], the predicted proteins in the whole genome of A. westerdijkiae cover all enzymes that digest cellulose, xyloglucan, xylan, galactomannan, starch, insulin and especially pectin (the biological function of which is to crosslink cellulose and hemicellulose fibres). This species is therefore likely to colonise and damage cotton canvas. Penicillium chrysogenum, formerly P. notatum and currently reidentified as P. rubens–mainly known for its use in producing the antibiotic penicillin [44]–is also a filamentous fungus that is very common in indoor air, dust and damp building materials [45]. Penicillium chrysogenum can proliferate at between 15 °C and 25 °C and a_w_ between 0.75 and 0.85, although preferably at a_w_ = 0.8, regardless of temperature [46]; it can also grow at a_w_ between 0.85 and 0.99, but not between 0.65 and 0.75 [47]. Similarly, Abellana et al. [48] observed that P. chrysogenum does not grow at a_w_ < 0.85 at temperatures between 20 °C and 30 °C, or at a_w_ < 0.9 at 15 °C. In flax canvas paintings exhibited in Slovenian churches and museums, P. chrysogenum was found to be the most abundant species of the genus Penicillium among those isolated and cultivated [49]. This species produces a wide range of enzymes, including amylase, lipase, protease, β-mannanase and hemicellulose [50], which can catalyse the hydrolytic decomposition of cellulose fibres, thus affecting the integrity of cotton canvas fabric.

Analysis of samples by FTIR-ATR spectroscopy enabled identification of the adhesives in the commercial tape strips Filmoplast^®^ P and Fungi-Tape^TM^ (Figure 1). The infrared spectrum of Filmoplast^®^ P is characteristic of an acrylic-based adhesive, with a strong peak at 1728.14 cm^−1^ (C=O stretching of ester groups) and typical bands at 1237.71 cm^−1^ (C-O stretching) and 1155.81 cm^−1^ (stretching vibration of ester group), although with bands also appearing at 1447.11 cm^−1^ and 1378.39 cm^−1^ (CH bending) and 2957.76 cm^−1^, 2929.51 cm^−1^ and 2872.27 (CH stretching), which overlap with the characteristic profile of cellulose, with bands between 2930 and 2850 cm^−1^ (CH stretching). Fungi-Tape^TM^ appears to be composed of a natural rubber, characterised by strong peaks at 2952.78 cm^−1^, 2917.79 cm^−1^, 2863.84 cm^−1^ (CH stretching), 1449.04 cm^−1^ and 1374.63 cm^−1^ (CH asymmetric deformation), 887.53 cm^−1^ (asymmetric stretching of C–O–C), 836.90 cm^−1^ (CH bending) and 748.23 cm^−1^ (CH rocking). According to the companies that produce pressure-sensitive adhesive tape, rubber-based adhesives have a higher initial tack than acrylic, thus enabling adherence to rough-textured material. Thus, Fungi-Tape^TM^ should adhere more strongly than Filmoplast^®^P. Figure 1 also shows that the two adhesive tapes did not leave appreciable residues on the canvas, after their removal the canvas did not show the typical FTIR peaks of the adhesives. This aspect is explained in further detail later on in this section.

Figure 2 shows FESEM images of PVA-B gels, both unmodified and functionalized with CEO or Cyrene™ (in the quantities finally chosen, as described below). The unmodified PVA-B gel shows a continuous and relatively homogeneous structure with a reticulated morphology, which can be attributed to the formation of a physically crosslinked polymeric network resulting from diol–borate complexation between hydroxyl-rich PVA chains and borate ions generated during the borax dissolution [24]. A similar overall structure is observed in PVA-B functionalized with CEO. The presence of cavities containing encapsulated spherical structures can be identified, which may be associated with the incorporation of the essential oil within the gel matrix. PVA-B gel functionalized with Cyrene™ displays a more heterogenous microstructure, suggesting that the inclusion of this solvent influences the organisation of the polymeric network during the gel formation. FESEM images of the PVA-B gel after used for fungal removal reveal the presence of entrapped spores within the gel matrix, supporting a physical cleaning mechanism based on the encapsulation and removal of fungal agents. No visible damage to the gel structure was observed after treatment, indicating that the cleaning process does not compromise the gel’s structural integrity.

The results of the three methods of applying the PVA-B gel (with/without CEO or Cyrene™) to the mock-ups with fungal growth are shown in Figure S3. The desired outcome was maximum removal of fungal colonisation with minimal gel residue on the surface. Applying and leaving the PVA-B gel for 15 min and 30 min before peeling it off removed small amounts of fungus, but large amounts of gel residue remained on the surface, trapping the fungal material between the PVA-B gel and the canvas. This is consistent with the findings of previous studies where PVA-B gel was also left in contact for some time with the material to be removed. Stagno et al. [51] observed gel residues, especially inside the large pores, after application of PVA–borax hydrogel for 4 min and 2 h to clean a Paraloid B72 coating from the surface of limestones (Travertine and Lecce Stone). Riedo et al. [52] conducted a study using canvas mock-ups with aged dammar varnish treated with PVA–borax gels, observing that all surfaces left in contact with the gel for 4 min had gel residues (identified by pyrolysis–gas chromatography/mass spectrometry, Py-GC/MS). These authors concluded that the surfaces had to be cleaned using cotton swabs and organic solvents to ensure the removal of gel residue. However, the method of continued dabbing used in the present study maximised the amount of pressure and the contact between the gel and the material to be removed. Thus, almost all of the fungal material was removed, except from the mock-ups with dense coverage of P. chrysogenum, and no gel residue remained on the surface. This method of application was therefore selected for the rest of the study. None of the three methods caused physical changes (fibrillation) to the surface of the canvas mock-ups.

The semi-quantitative results of the fungal inhibition rate are shown in Figure S4. For A. westerdijkiae, PVA-B gel containing CEO at concentrations of 0.50% and 1.00% yielded low-to-medium fungal inhibition (halo radius, 0.1–1.0 cm), while at 1.50% the inhibition was medium-to-high (halo radius 1.3–2.5 cm). For P. chrysogenum, the PVA-B gel containing CEO at a concentration of 0.50% yielded high inhibition of fungus (halo radius, 1.0–1.8 cm), while concentrations of 1.00% and 1.50% yielded complete inhibition. The PVA-B gel containing 16% of Cyrene™ yielded low-to-medium fungal inhibition (halo radius, 0.0–0.3 cm) for A. westerdijkiae, while a high level of inhibition (halo radius 1.0–2.2 cm) was obtained with P. chrysogenum. The PVA-B gel containing 8% and 12% of Cyrene™ yielded low inhibition (halo radius, 0.0–0.9 cm) of P. chrysogenum. Based on these results, a mixture of 1.5% CEO and 16% Cyrene™ was chosen for separate incorporation in the PVA-B gel matrix.

The chemical characterisation of CEO and Cyrene™ by FTIR and GC-Orbitrap-HRMS is shown in Figure 3. According to this analysis, the main chemical components of the CEO were eugenol (93.99%) and caryophyllene (3.42%). The FTIR spectrum showed that CEO had a similar vibration structure to eugenol in the fingerprint region of 1500 cm^−1^–500 cm^−1^, with a strong peak at 1513.12 cm^−1^ (stretching vibration of C=C) and several peaks in the range of 1267.86 cm^−1^ to 1100 cm^−1^ (asymmetric stretching of C–O–C). Other characteristic peaks of eugenol are shown, such as a broad peak at 3422.59 cm^−1^ (OH stretching), and peaks at 3073.38 cm^−1^, 2935.48 cm^−1^ and 2840.38 cm^−1^ (CH stretching). Peaks at 1431.37 cm^−1^ and 877.18 cm^−1^ were caused by the presence of the CH_2_ group, and a peak at 1762.19 cm^−1^ may be derived from a carbonyl ester attached to the aromatic ring of eugenol acetate, characteristic of CEO. Dihydrolevoglucosenone (98.99%) was the only compound identified in Cyrene™ by GC-Orbitrap-HRMS analysis. Cyrene™ FTIR spectrum matches published spectra of the compound [53], with peaks at 2967.05 cm^−1^ and 2901.36 cm^−1^ (CH stretching), 1741.82 cm^−1^ (C=O stretching), 1418.72 cm^−1^ (C−H stretching), 1285.72 cm^−1^, 1244.63 cm^−1^ and 1209.97 cm^−1^ (C-O stretching), and a strong peak at 1110.54 cm^−1^ (-C-O-C- stretching). Likewise, the FTIR spectra of the PVA-B gel matrix alone and containing each of the compounds are shown in Figure 4. In all three cases, peaks appeared at, respectively, 3254.18 cm^−1^, 3246.26 cm^−1^ and 3261.46 cm^−1^ (OH stretching), characteristic of the water-based matrix of PVA-B gel, peaks at 1286.90 cm^−1^ and 1287.00 cm^−1^ and 1094.00 cm^−1^, 1108.12 cm^−1^ and 1108.85 cm^−1^, corresponding to the stretching vibrations of B-O-C bonds, peaks with low intensity at 1413.64 cm^−1^, 1430.19 cm^−1^ and 1439.32 cm^−1^, corresponding to the stretching vibration of B-O bonds, and peaks at 1635.55 cm^−1^, 1636.65 cm^−1^, 1637.46 cm^−1^, corresponding to the stretching vibration of C=O that is assigned to the boric ester bond, which confirmed that the material is a PVA-B gel. The bands corresponding to the potential antifungal agents (CEO and Cyrene™) tested were scarcely visible. In the PVA-B gel+ CEO spectrum, a very slight peak was observed at 1513.16 cm^−1^, which appeared as an intense peak at 1513.12 cm^−1^ in the FTIR spectrum of CEO (Figure 3). In the case of PVA-B gel+ Cyrene™, the corresponding signal was slightly more noticeable, probably due to the greater amount of this compound used, and it appeared at 984.91 cm^−1^, in the Cyrene™ spectrum (Figure 3) being at 985.84 cm^−1^.

Micro-aspiration, Filmoplast^®^ P, Fungi-Tape™, PVA-B gel, PVA-B gel + Cyrene™ 16% and PVA-B gel+ CEO 1.5% were used to remove A. westerdijkiae and P. chrysogenum from cotton mock-ups. The treatments were applied until complete removal (no visible fungal residues) or when no further fungal residue could be eliminated. For micro-aspiration, the maximum application time required was 10 s for sparse coverage of A. westerdijkiae (AS) and sparse coverage of P. chrysogenum (PS), 30 s for dense coverage of A. westerdijkiae (AD) and 120 s for dense coverage of P. chrysogenum (PD). For Filmoplast^®^ P, the maximum number of adhesive tape strips used was 8 strips for AS, 4 for PL, 8 for AD and 16 for PD. Similarly, for Fungi-Tape™, 8 strips were required for AS, 3 for PL, 6 for AD and 15 for PD. For PVA-B gel, the number of times the mock-up was dabbed with the gel to remove the fungal colonisation was recorded. In mock-ups with sparse coverage, dabbing 10 times was sufficient to remove the fungal growth, while for AD and PD, the surfaces had to be dabbed 15 times and 20 times, respectively. The findings demonstrate a link between the degree of coverage and the maximum number of treatment applications needed. With PVA-B + Cyrene™ 16%, dabbing 9 times removed sparse coverage, while the mock-ups had to be dabbed 13 times to remove AD and 18 times to remove PD. With PVA-B + CEO 1.5%, dabbing 10 times was sufficient to remove sparse coverage, while AD was removed by dabbing 15 times and PD by dabbing 20 times.

Figure 5 and Figure 6 show the removal capacity and fibrillation produced by the treatments evaluated by DOM and FESEM. After micro-aspiration, aerial and vegetative mycelia were observed on the cotton mock-ups. The presence of aerial mycelium residues was notable on the surface of the PD mock-ups and in the cotton fibres, together with slight fibrillation. Although the adhesive tape strips were the most effective for removal of the fungal colonisation, FESEM images revealed fungal residues within the mock-up fibres and the presence of some spores on the fibres, particularly in the PD mock-ups. The tape strips also produced a higher degree of fibrillation than the other treatments. The fibres in the mock-ups treated with Fungi-Tape™ were more affected than those in Filmoplast^®^ P in which the degree of fibrillation was lower (although still high). This may be explained by the chemical composition of both adhesives, since Fungi-Tape^TM^ with rubber-based adhesive has a higher tack than Filmoplast^®^ with an acrylic-based adhesive (Figure 2). The PVA-B gel was able to remove the fungal colonisation from AS, AD and PS mock-ups, and it was most effective for removing fungal material from inside the cotton fibres. By contrast, the results for PD were similar to those obtained by micro-aspiration, with large amounts of aerial mycelium residues remaining on the surface. Furthermore, this was the only treatment that did not cause alterations in the fibres of any of the mock-ups tested. No adhesive or gel residues were detected in any case. Penicillum chrysogenum has a complex, branched conidiophone structure, which may favour better adherence of hyphae to surfaces, with mechanical removal thus being more difficult. By contrast, A. westerdijkiae has a simple, unbranched conidiophore, which may facilitate its detachment. This may explain why removal of A. westerdijkiae is easier than of P. chrysogenum. In general, the degree of fibrillation caused by all treatments on the cotton canvas was low, as can also be seen in Figure S5, which meant that performance mechanical tests were not necessary. Furthermore, since the study is based on the conservation of heritage paintings and similar objects, the visual appearance for the preservation of the piece is the main evaluation criterium.

The results of ImageJ analysis (Figure 7) show that for sparse coverage, PVA-B was the most effective treatment, followed by Fungi-Tape™ and Filmoplast^®^ P, which removed similar amounts of the fungal colonisation. Micro-aspiration was slightly less effective than the adhesive tapes. In general, all treatments worked well in removing the fungi, although P. chrysogenum proved more difficult to remove than A. westerdijkiae. This may be due to the greenish colour of the former, which more visible on the cotton than the latter, which is a yellowish colour, similar to the background. All treatments were less effective in removing dense coverage of fungus than sparse coverage. PVA-B was more effective than the other treatments for removing the dense coverage of A. westerdijkiae. Both adhesive tapes yielded similar results, poorer than that of PVA-B. Nonetheless, Filmoplast^®^ P was capable of removing more of the fungal colonisation than Fungi-Tape™, which yielded more variable results. The variability may be due to the high adhesive strength of Fungi-Tape™, which prevents good control of the treatment application relative to Filmoplast^®^ P, and thus more variable results. By contrast, for the dense P. chrysogenum coverage, adhesive tapes were the most effective, with Filmoplast^®^ P again producing better results than Fungi-Tape™. Micro-aspiration and PVA-B yielded similar removal results, although micro-aspiration removed slightly more of the fungal colonisation. The high variation in the effects of micro-aspiration in the case of the two fungi may also be due to the lack of good control in applying the treatment. All treatments eliminated around 50% of the fungal colonisation.

ATR-FTIR spectra of the canvas mock-ups after applying Fungi-Tape™ and Filmoplast^®^ P are shown in Figure 2, even after applying PVA-B gel, PVA-B gel+ CEO (1.5%) and PVA-B gel + Cyrene ^TM^ (16%) are shown in Figure 4. Peaks corresponding to the remains of adhesive or gel were not observed in any case. Only peaks for cotton fibres and typical bands of cellulose, lignin and hemicellulose were observed, such as the strong peak observed at 3305.36 cm^−1^ (PVA-B gel), 3311.15 cm^−1^ (PVA-B gel+ CEO 1.5%) and 3284.15 cm^−1^ (PVA-B gel + Cyrene ^TM^ 16%), as well as 3274.18 cm^−1^ (Filmoplast^®^ P) and 3268.38 cm^−1^ (Fungi-Tape™), which is characteristic of the hydroxyl (OH) groups of cellulose, lignin and water. The peaks at 2888.82 cm^−1^ (PVA-B gel), 2902.32 cm^−1^ (PVA-B gel+ CEO 1.5%), 2865.68 cm^−1^ (PVA-B gel+ Cyrene ^TM^ 16%), 2899.87 cm^−1^ (Filmoplast^®^ P) and 2895.70 cm^−1^ (Fungi-Tape™) are characteristic of the stretching vibration of C-H present in cellulose and hemicellulose. The band at 1633.40 cm^−1^, 1627.61 cm^−1^, 1643.04 cm^−1^, 1641.63 cm^−1^ and 1652.65 cm^−1^, respectively, according to the above order, may be related to the presence of water in the fibres. Intense peaks at 1022.08 cm^−1^, 1018.22 cm^−1^, 1022.08 cm^−1^, 1029.72 cm^−1^ and 1052.23 cm^−1^ are related to the CO and OH stretching vibrations of the polysaccharide in cellulose. The peaks at 885.16 cm^−1^, 885.16 cm^−1^, 890.94 cm^−1^, 885.16 cm^−1^ and 894.48 cm^−1^ are related to the presence of β-glycosidic linkages between monosaccharides.

The antifungal activity of PVA-B gel with CEO 1.50% and Cyrene™ 16% is shown in Figure 8. Initial assays showed that CEO inhibited the growth of A. westerdijkiae and P. chrysogenum immediately after fungal removal (0 h) and also at 24 h, while Cyrene™ only inhibited growth of A. westerdijkiae 24 h after fungal removal (24 h) (Figure 8A). In vitro tests, performed with the remains (collected with sterile swabs) on colonised mock-ups treated with PVA-B gels with 1.5% CEO and 16% Cyrene™, revealed active fungal residues on the mock-ups (Figure 8B). The preliminary study of the antifungal activity by confocal microscopy indicated that most of the fungal cells (58–68%) remaining inside the canvas fibres were damaged. They may be repaired, and remaining live cells are potentially dangerous. Around 25% remained alive, except for A. westerdijkiae treated with PVA-B gel+ Cyrene™, in which the proportion of live cells was reduced to 12% (Figure 8C). Viability staining (based on membrane integrity) does not clearly differentiate sub-lethal damage and is unable to separate between cells that will recover and those that will not. To overcome this limitation, ATP assays could have been used as complementary methods that assess different aspects of cell health (vitality) beyond only membrane integrity.

Finally, the potential benefits and drawbacks of the six treatments tested for the removal of biodeteriogenic fungal strains from cotton canvas were identified and summarised in Table 1 using a comparative approach. In addition to the aspects discussed in the paper, an assessment of the toxicity and cost of each treatment were included.

3. Conclusions

This preliminary study evaluated the use of distinct types of cleaning methods, including micro-aspiration, adhesive tapes, and PVA-B gel with or without CEO and Cyrene™, to remove fungal colonisation from canvas paintings supports. The proposed methods were applied by avoiding, as far as possible, damaging the canvas fibres, ultimately for the conservation of heritage paintings and similar objects. The findings provide useful data on the performance of commonly used and emerging cleaning tools. The six different cleaning methods were assessed to determine which (if any) could achieve the ultimate goal: the efficient removal of fungal material without damaging the artwork support. Aspergillus westerdijkiae was easier to remove than Penicillium chrysogenum, especially with dense fungal coverage. In general, adhesive tape strips removed more fungus than micro-aspiration but less than PVA-B gel with or without CEO and Cyrene™. Both tape strips failed to remove the parts of the fungi that had penetrated and found on the subsurface of the canvas fabric, while fungi on the subsurface did were achieved by PVA-B gel (with and without CEO and Cyrene™) without damaging the fibres. Filmoplast^®^ P was most effective treatment for removing dense coverage of Penicillium chrysogenum, the fungus that was the most difficult to remove. This tape also caused less fibrillation than the Fungi-Tape^TM^, which adhered more strongly to the cotton canvas. Regarding the antifungal activity, the preliminary tests showed that PVA-B gel+ Cyrene™ performed similarly to PVA-B gel+ CEO; however, further research is needed to completely confirm the efficacy of the treatments.

4. Materials and Methods

4.1. Cotton Canvas Mock-Ups with Fungal Colonisation

Cotton textile (density: 12 × 15 threads/cm^2^) was acquired from Art i Clar (Valencia, Spain) to prepare the canvas mock-ups of 4 cm x 5 cm in weft × warp direction with fungal colonisation (Figure 9a). Cotton textile was selected because it is commonly used as the fabric support for canvas paintings, it is hygroscopic, and also because the high cellulose content promotes fungal growth. The mock-ups were sterilised with formaldehyde solution 37% v/v (Scharlab^®^, Barcelona, Spain), following the protocol described by Sanmartín et al. [54]. To ensure complete evaporation of formaldehyde and prevent the presence of residues at the time of fungal inoculation, the samples were held in a laminar flow cabinet (Cruma 870 fl) for 48 h. The sterilised mock-ups were placed on Sabouraud dextrose agar (SDA) culture medium in Petri dishes and inoculated separately with two suspensions of Aspergillus sp. and Penicillium sp. (300 µL, 10^6^ CFU/mL, after both strains reached an optimal level of growth for inoculation with active sporulating conidiophores). For inoculation, 6 drops (each approx. 50 µL) were deposited at evenly distributed points across the entire surface area of 20 cm^2^. The drops were then spread out evenly with a triangular loop (Drigalski spreader). The mock-ups were incubated at 28 °C (in a Binder^®^ BD 115 incubator) for between 15 and 21 days. The underlying SDA medium provided the fungus in the fabric sufficient moisture and the nutrients necessary for accelerated growth. Thus, the fungus developed much earlier on the agar medium and took between 2 and 3 weeks to grow on the mock-up; the fungal colonies were clearly observed, even at low coverage. In this respect, two degrees of colonisation, i.e., sparse and dense, were established on the basis of the area covered by the fungus. For dense coverage (D), more than 50% of the surface area of the canvas mock-up (20 cm^2^) was covered by the fungus, while for sparse coverage (S), less than 50% of the surface area was covered by the fungus (Figure 9a).

Once the colonised mock-up replicas were prepared, they were stored in sterile Petri dishes without culture medium and held in the laboratory (40–45% RH, 20–25 °C) for 2–3 months before application of the test methods. The aim of this second incubation period was to emulate the brief natural process that follows the outbreak of massive fungal colonisation. The agar medium was removed for that purpose, intended to stabilise colonised samples rather than promote further growth because fungal growth is limited at RH values below 80%. Four canvas mock-ups were selected for each fungus and degree of coverage: Penicillium sp. dense coverage (PD), Penicillium sp. sparse coverage (PS), Aspergillus sp. dense coverage (AD), and Aspergillus sp. sparse coverage (AS). Three of the four canvas mock-ups were cut into four pieces, to produce twelve replicates each measuring 4 cm x 1.25 cm. The replicates were randomly distributed in groups of three for testing the efficacy of each of the four removal methods: micro-aspiration, Fungi-Tape^TM^ and Filmoplast^®^ P adhesive tape strips, and PVA-B gel. The remaining mock-up was also cut into four pieces, which were used to determine the best way of applying the PVA-B gel (Figure 9b, Section 4.4). Similarly, for testing the PVA-B gel containing antifungal agents, four mock-ups were prepared for each fungus and degree of coverage and then cut into four pieces of 4 cm × 1.25 cm. Of the sixteen replicates obtained, ten were reserved for additional experiments, such as the application test of PVA-B gel with CEO or Cyrene™ embedded. The six samples chosen at random were used for the removal with PVA-B gel+ CEO (n = 3 samples) and PVA-B gel + Cyrene™ (n = 3 samples) after determining the amount of CEO and Cyrene™ to be included in the medium (Figure 9b, Section 4.4).

4.2. Identification of Fungi

The tested fungal strains were obtained from a canvas painting attacked by fungi (Section 1) and selected as putative strains of Aspergillus (sequence ID728-T6SABB) and Penicillium (sequence PR1) based on their morphological characteristics. Genomic DNA was purified from pure colonies and fungal rDNA genes containing the ITS and D1/D2 regions were amplified by PCR with an ITS5/LR6 set of primers (by STABVida^®^ Lda. Caparica, Portugal) for the ID728-T6SABB sequence. For the PR1 sequence, the β-Tubulin genomic region was sequenced and analysed using the primers Bt2a and Bt2b (by CECT, Valencia, Spain).

The polymerase chain reaction (PCR) conditions for sequence ID728-T6SABB consisted of one cycle at 95 °C for 30 s, followed by 40 cycles of denaturation at 95 °C for 15 s, annealing at 52 °C for 15 s, extension at 72 °C for 15 s, and a final extension at 72 °C for 2 min. The PCR conditions for the sequence PR1 consisted of one cycle at 95 °C for 5 min, followed by 35 cycles of denaturation at 94 °C for 45 s, annealing at 55 °C for 45 s, extension at 72 °C for 1 min and a final extension at 72 °C for 10 min.

Fungal sequences were compared against the NCBI public reference database using BLAST (Basic Local Alignment Search Tool) analysis. The sequences obtained were aligned with those retrieved from the GenBank database (http://www.ncbi.nlm.nih.gov/, accessed on 31 January 2026) using the BioEdit v.7.7.1 software, in order to identify the species. The constructed trees (created using FigTree v.1.4.4) are presented in Section 3. The sequences were deposited in the GenBank NIH genetic sequence database (GenBank: https://www.ncbi.nlm.nih.gov/genbank/, accessed on 31 January 2026) under accession numbers PX315808 (sequence ID728-T6SABB) and PV792917 (sequence PR1).

4.3. Micro-Aspiration and Adhesive Tape Strips

A Muntz Museum Vacuum Cleaner 555-MU-E HEPA with rigid nozzle PHU-14 (CTS, Getafe, Spain) and a power output of 1300 watts (Figure 9b) was used for micro-aspiration. Fungi-Tape™ (Quima, Valencia, Spain) transparent cellulose strips (360 units) of 6.6 cm × 2.2 cm, and Filmoplast^®^ P (CTS, Getafe, Spain) transparent Japanese paper (20 g/m^2^) tape (1 unit) of 500 cm × 2 cm, were used as adhesive tape strips. In both cases, pieces of strips measuring 6 cm × 2 cm were used. The adhesive side of both tapes was analysed by attenuated total reflectance Fourier transform infrared spectroscopy (ART-FTIR). The ATR-FTIR spectra were obtained using a Fourier Vertex 70 spectrometer (Bruker Optik GmbH, Bremen, Germany), equipped with a FR-DTGS (fast response, thermally stabilised deuterated triglycine) detector and a Golden Gate MKII attenuated total reflectance accessory. Scanning of each tape was conducted at a spectral resolution of 4 cm^−1^. The spectra were processed using OPUS 5.0/IR software (Bruker Optik GmbH, Germany).

4.4. PVA-B Gel (With/Without CEO or Cyrene™): Formulation and Application Trials

PVA-B gel was prepared using polyvinyl alcohol (PVA) (67,000 Mw, 87–89% hydrolysis), formula (C_2_H_4_O)x, and anhydrous disodium tetraborate (borax), formula Na_2_B_4_O_7_, (CTS, Getafe, Spain). Both were separately dissolved in distilled water at a concentration of 8% (w/v), with the aid of a heating magnetic stirrer (VELP Scientifica, model ARE, Milano, Italy), at 90 °C for approximately 1 h (PVA) and 15 min (borax), with constant magnetic stirring, until homogeneous mixtures were obtained. The dissolutions were then mixed at a 4:1 (v/v) ratio (PVA:Borax) following the protocol described by Altobelli et al. [22], resulting in a final concentration of 6.4% (w/w) PVA and 1.6% (w/w) borax in the gel.

The method of applying the PVA-B gel (with/without CEO or Cyrene™) to the mock-ups, to achieve maximum fungus removal and minimal gel residue, was also tested. Three different methods of application were tested: (1) pressing the PVA-B gel on the mock-up for 15 min before removal by peeling at a 90-degree angle (ASTM D3330, 2025 [55]), (2) the same as above but with a contact time of 30 min, and (3) continued dabbing of the mock-up with the PVA-B gel (Video S1, Supplementary Material). The results were evaluated relative to untreated samples using digital optical microscopy (DOM) Nikon SMZ1500 (Nikon, Tokyo, Japan) with a digital videomicroscope Dino-Lite AM4113 T-FVW, CTS (Getafe, Spain) and field emission scanning electron microscopy (FESEM) (Ultra Plus Zeiss with EDX Quorum Q150T S plus, Zeiss, Jena, Germany) at 3 Kv. The microstructural morphology of the PVA-B gels, both unmodified and functionalised with CEO or Cyrene™, was characterised by CryoFESEM. Gel samples were cryo-fixed by rapid immersion in liquid nitrogen and freeze-dried by controlled sublimation for 15 min at −90 °C using Quorum Technologies PP3010T cryo-preparation system to preserve their microstructural morphology prior to FESEM analysis using the equipment previously described. Micrographs of unmodified PVA-B gels after fungal removal were obtained using the same FESEM equipment.

Formulation tests were carried out by increasing the amounts the antifungal agent used, to determine the effective minimum amount of the clove (Eugenia caryophyllus) essential oil (CEO) (Pranarom^®^, Barcelona, Spain) and the bio-based, non-toxic and biodegradable solvent Cyrene™, dihydrolevoglucosenone (Sigma-Aldrich^®^, Darmstadt, Germany). For CEO, the PVA matrix (before adding borax) was mixed (using the same magnetic stirrer indicated above) at room temperature with 2% Tween^®^ 20 (Sigma-Aldrich^®^, Darmstadt, Germany), a nonionic surfactant added to achieve miscibility of the oil in the water-based PVA gel matrix [56], and five different concentrations of CEO (0.03%, 0.10%, 0.50%, 1.00% and 1.50%) [57] were then added. Borax was added at the same ratio regarding the PVA indicated above and mixed at 90 °C with a glass rod. For Cyrene™, different amounts (1%, 5%, 8%, 12% and 16%) [58] were separately added and mixed in the same way as CEO but without the addition of Tween^®^ 20. Note that the maximum amount of Cyrene™ that could be incorporated into the PVA gel matrix was 16%. PVA-B gel alone and with 2% Tween^®^ 20 were used as controls.

The antifungal activity was evaluated semi-quantitatively by the Kirby–Bauer test or disc diffusion test assays, in order to determine the lowest concentration of the potential antifungal agent (CEO or Cyrene™) required to inhibit visible microbial growth in vitro. Hence, Petri dishes containing SDA medium were separately inoculated with 100 µL (10^6^ CFU/mL) of both fungal suspensions under the same conditions described in Section 4.1. Three equidistant 0.9 cm circular holes were made in the SDA medium in the Petri dishes, with a sterile metal punch, and a suitable volume (250–320 µL) of the gel solution containing CEO or Cyrene™ at the chosen concentration was added to the well. The plates were incubated at 28 °C for 5 days, and the inhibition halo around each fungal colony was measured. The experiments were performed in triplicate. To evaluate the fungal inhibition rate, the scale proposed by Gatti et al. [34] was used to determine the degree of inhibition: moderate inhibition (0.2–1 cm halo), high inhibition (1–2 cm halo), total inhibition (absence of fungal growth) and no inhibition (complete fungal growth).

The compositions of CEO and Cyrene^TM^ were determined by gas chromatography and Orbitrap high-resolution mass spectrometry (GC-Orbitrap-HRMS). Both samples were diluted with dichloromethane (CH_2_Cl_2_) and analysed directly. To enhance the accuracy and reliability of compound identification, retention indices were used for correction alongside the mass spectrometry data. Fourier transform infrared (FTIR) spectroscopy was also used, under the same device and conditions described in Section 4.3, to analyse both antifungal agents with the KBr pellet method. The PVA-B gel, PVA-B gel+ CEO at 1.50% (selected amount) and PVA-B gel+ Cyrene™ at 16% (selected amount) were analysed in ATR mode.

4.5. Removal of Fingal Colonisation from Cotton Canvas

All treatments were applied until all visible fungal growth had disappeared or no further fungal colonisation could be removed. To prevent contamination, all treatments were carried out in a Cruma 870-FL vertical laminar flow hood, with the contaminated mock-ups placed in empty sterile Petri dishes. To facilitate removal, sterile metal forceps were used to hold the mock-ups, especially during micro-aspiration (Figure 9b). For micro-aspiration, the application time (in seconds) was measured. For Fungi-Tape™ and Filmoplast^®^ P tapes, the number of strips used was counted. The strips were pressed on to the entire surface of the colonised mock-ups with sterile cotton swabs and were then removed by peeling at a 90-degree angle (ASTM D3330, 2025 [55]). For PVA-B gel (with/without CEO or Cyrene™), the number of times the mock-up was dabbed with the gel (selected method of application) was counted.

To assess the removal of the fungal growth from the canvas and the possible fibrillation and remaining adhesive (in the case of strips) or gel (in the case of PVA-B gel), DOM and FESEM images were captured after treatment, along with images of untreated samples, using the same device and conditions described in Section 4.4. In addition, the efficacy of removal (as the percentage fungal colonisation removed) was evaluated using ImageJ software v.1.54. with the Trainable Weka Segmentation plugin. Photographs were taken in random areas of the mock-ups with fungal contamination using a digital videomicroscope (Dino-Lite AM4113 T-FVW, CTS, Getafe, Spain), with three replicates for each fungal strain, degree of coverage and removal method (24 images in total). The photographs covered a surface around 41.45 mm^2^.

The data were examined by Student’s t-test (p-value ≤ 0.05) and ANOVA, with a Duncan post hoc test (p-value ≤ 0.05), using RStudio v. 12.0 software. Any remains of adhesive or gel on the canvas mock-ups after applying Fungi-Tape™, Filmoplast^®^ P, PVA-B gel, PVA-B gel+ CEO (1.5%) or PVA-B gel + Cyrene ^TM^ (16%) were analysed by FTIR-ATR using the same device and conditions described in Section 4.3.

4.6. PVA-B Gel with CEO or Cyrene™: Potential Antifungal Agents

To complement the previous antifungal assays (Section 4.4), antifungal activity assays were separately performed with the maximum and selected amounts of CEO and Cyrene™, 1.50% and 16%, respectively, incorporated in the PVA-B gel. These assays were conducted to assess whether the treatments can be considered with antifungal activity (Figure 9c). Thus, some of the samples of PVA-B gel+ CEO (1.5%) and PVA-B gel+ Cyrene ^TM^ (16%) used for fungal removal from cotton canvas (Section 4.5) were placed in the holes punched in Petri dishes, as in the previous antifungal assays (Section 4.4). Two groups of samples were placed in the holes: PVA-B gel with CEO or Cyrene™ immediately after fungal removal (0 h) and PVA-B gel with CEO or Cyrene™ 24 h after fungal removal (24 h) (the samples were stored under sterile conditions during this time). The plates were incubated at 28 °C, and the fungal growth was checked after 5 days. The experiments were performed in triplicate.

After removal of the fungal growth (Section 4.5), the cotton canvases were used separately for (1) in vitro tests (in triplicate), in which viable fungal remains on the surface canvas were collected using sterile cotton swabs and incubated in SDA plates for 5 days at 28 °C to check whether fungal growth occurred (Figure 9c), and (2) cell staining and identification of live, damaged and dead fungal cells by confocal laser scanning microscopy (CLSM) (Figure 9c) (Leica TCS SP5 X AOBS, Leica Microsystems Heidelberg GmbH, Mannheim, Germany) equipped with white light laser (WLL) and acousto-optical-beam-splitter (AOBS^®^). The confocal images were acquired in fluorescence mode using a HCX PL APO CS 20× 0.7 dry objective lens with a 3× zoom. A defined area of the canvas (1 cm × 1 cm) was stained with 3 µL of SYTO^®^ 9 and 3 µL of propidium iodide from LIVE/DEAD^®^ FungalLight™ Yeast Viability Kit (Thermo Fisher Scientific, Waltham, MA, USA), 15–30 min prior to the analyses. The samples were excited by two laser lines, the 480 nm and 640 nm, and the corresponding emission spectra were recorded at 490 nm–510 nm for SYTO^®^ and 650 nm–701 nm for propidium iodide. Counting of live (green), damaged (orange) and dead (red) fungal cells was carried out using ImageJ software (version 1.54e). These experiments were not replicated.