1The earliest known mention of the greening of Oxford stone was by W.J. Arkell (1947) in his book Oxford Stone. In his Chapter 8 on The Decay, Repair and Maintenance of Oxford Buildings, he referred to lichens and algae, including what he considered to be the most characteristic plants, namely the bright green alga, Pleurococcus naegelii Chodat (= Protococcus viridis Agardh). He stated that “[i]n winter it enlivens with its bright green mantle many a dismal north wall”, including the north side of St Swithun’s and the President’s Lodgings at Magdalen College as well as along the front of New College’s New Buildings that front onto Holywell Street (Arkell, 1947, p. 160). He argued that even though Pleurococcus can be seen winter after winter it does not harm the stone, which is more weathered on the south side of buildings that are exposed to sunlight and rainfall.
2J.J. Ortega-Calvo et al. (1995) more recently described Pleurococcus as a ubiquitous alga that is found in various humid terrestrial substrata, where algal development on building stones often occurs in places of high humidity and water retention with very low light intensity. These researchers also noted that at the Pardon Gate of the Cathedral of Seville, Spain seasonal observations during summer and repeated in winter following a period of heavy rainfall saw the same range of species present in summer despite intense algal growth in winter, suggesting algal preservation in drought periods when temperatures are high during summer. They also discovered that the length of period of wetness as well as aspect (orientation, or direction of exposure often governed by architectural structure, e.g. cornice, crevice, or vertical wall) affect algal biofilm formation, depending on whether the stone remains wet for a sufficient length of time for the production of organic matter as well as protosoil development and the establishment of mosses and higher plants. South-facing rocks of the San Millán church, Spain were colonised by the lichen Physcia dimidiata and accompanied by mosses, especially in areas of rising damp and on horizontal surfaces, which were blackened due to lichen growth as well as dematiaceous fungi, including Phialophora sp. and Cladosporium tenuissimun (de los Rios et al., 2009). Elsewhere, green algae (Chlorococcum and Muriella) were found in association with mosses and lichens in early development (Nugari et al., 2009), as for example brilliant green patinas appearing in moist areas of mural paintings. These workers suggested, rather than focusing on ventilation, that humidity and sunlight exposure should be reduced as preventive interventions at the Crypt of the Original Sin in Matera, Italy. Research by M.J. Duane et al. (2003) at Mimoosa Beach near Mohammedia in northwest Morocco also identified light as an important controlling factor for boring algae.
3Sunlight promotes the development of photosynthetic organisms on stone surfaces, and this growth is also enhanced by the availability of moisture in the environment. Wetting can establish acidity through the formation of acids, but (decomposing) plants also release acids, which can etch surfaces. For example, some laboratory studies were conducted by R.G. Welton et al. (2003), which revealed the etching of minerals by axenic algal populations within just 90 days. The algae were also found to have a significant effect on chemistry in vicinity of rock chips and to absorb Ca in particular, but also Mg ions, from surrounding liquid. E. Uchida et al. (1999) mentioned that Ca and Mg are needed for gypsum formation, which can be derived from stones (such as sandstones). A.Z. Miller et al. (2008) performed laboratory studies in which the growth of a phototrophic community was visible after just one week of inoculation, with a bright green colouration that after one month became a dense green colour and continued that way for up to 3 months. After this period, white colonies (fungi) appeared on the green biofilms that covered stone surfaces. Algae (and cyanobacteria), as a photosynthetic microorganism, establish first before heterotrophic microorganisms, such as bacteria and fungi. G.M. Gadd (2007) outlined the colonisation of fresh stone as comprising in the first and second year surface penetration by algae and ascomycetes followed by the establishment in the substratum of lichen-forming fungi, which takes several years. This finding had also been evident in research by M. Hoppert et al. (2004), who associated algal colonisation within the first two years of exposure to the environment, whereas colonisation by lichens occurred over the span of several years.
4Other researchers have investigated stone weathering through quantitative assessments, including chromatic change. Visible light cameras are now being employed as part of new methodologies, as in the field-based measurement of budburst and leaf area expansion (Graham et al., 2009). R.J. Hintz and K. Joon-Mook (1985) used cameras in a photogrammetric approach to measure and record historical art and architecture in the Republic of Korea. Others, such as L.B. Preskitt et al. (2004), have used a photoquadrat method employing a digital camera with computer software for photographic analysis and compared it with a more conventional method of grid quadrats to estimate percentage cover in subtidal benthic communities. The photoquadrat approach was more precise in estimating percentage cover of abundant species and provided a permanent visual record, although it did not produce results with a finer resolution in diversity found using point-intersect quadrats. Nevertheless, the two methods demonstrated a high level of similarity, particularly when taken on relatively flat surfaces (but not in crevices and holes on the reef).
5Integrated digital photography and image processing (IDIP) was originally developed by M.J. Thornbush and H.A. Viles (2004a) and first applied by the authors to quantify soiling patterns on stone sensor discs in the laboratory (Thornbush and Viles, 2004b). An outdoor version of the method was developed (namely, O-IDIP) and was used to quantify the exterior soiling of buildings at the façade scale (Thornbush, 2010a) as well as closer up scales (Thornbush and Viles, 2008), such as along a boundary wall (Thornbush and Viles, 2007) in the development of a related (digital mapping in Adobe Photoshop, or DMAP) method. O-IDIP has most recently been employed as a rephotographic technique to quantify outdoor weathering on building exteriors (Thornbush, 2010b). However, to-date, an emphasis has not be given to any particular colour channel, such as green-red (a) in order to measure biological colonisation of surfaces, as through the accumulation of algae. In the current study, the O-IDIP method is used to quantify the greening of walls in central Oxford, UK through the use of calibrated digital photography. This study provides further application of O-IDIP almost at the façade-scale of walls, but of a non-case-study of more than a single building, with an emphasis of colour change (discolouration?) of stone surfaces as an indicator of biodegradation and possibly biodeterioration. In this outdoor study, it is important to control microclimate in order to compare greening at different sites, such as of north-facing sites. However, because it was not completely possible to control microclimate in this (outdoor) environment, the results are used only to further illustrate application of the method with calibration, this time to greening.
6A high-resolution digital camera, namely a Nikon Coolpix S4, was used on 9 March 2010 in overcast conditions for (winter) photographic monitoring of north-facing walls along the south side of Broad Street in central Oxford, UK. Thirteen field sites were selected along this wide street in the Oxford city centre, where local effect of microclimatic impacts (associated with aspect) could be deduced more easily due the broadness of the street. Ten sites were located on the south side of the street (north-facing sites, including Sites 3-12), with three additional sites used as control sites located on the north side of the street (on south-facing walls, including Sites 1-2 and 13). This selection comprises mainly the north side of buildings, including Exeter College, the Museum of the History of Science, and the Sheldonian Theatre, but also includes the south side of buildings as control sites at Balliol and Trinity Colleges and the New Bodleian Library along this street. Digital photographs were taken approximately 1 m away from (and perpendicular to) wall surfaces and the ground. The digital camera was set to flash off, with macro on, and was completely zoomed out.
7The Nikon Coolpix S4 digital camera is a point-and-shoot type of compact camera with a maximum image resolution of 2816 x 2112 and 6.0 million effective pixels. Its pixel density is 24 MP/cm2. This is a 38-mm digital camera with swivel lens style, a high lens zoom of 10x, and a high digital zoom of 4x. The camera outputs JPGs in RGB Color Mode with a resolution of 300 dpi. These images were converted (undergoing a transformation in Mode) in Adobe Photoshop to TIFF in Lab Color Mode. Photoshop drops the asterisks typically used for Lab (L*a*b*) and employs a lightness (L) component, ranging from 0 to 100%. The chromatic components a (along the green-red axis) and b (blue-yellow axis) with values also ranging from 0 to 100%, when converted to a proportion from the raw values of 0 to 255 outputted by Photoshop for Lab Color.
8A calibration procedure was performed for each image similar to that previously outlined by M. Thornbush (2008). More specifically, calibration was conducted only based on a, as this was the channel focus of this study. The Color Sampler Tool was employed in Adobe Photoshop to sample No. 14 (green) on the ColorChecker chart, included in one of two images taken at each site (fig. 1A). One sample (#2, representing No. 14 for green on the chart) was used in order to simplify the calibration procedure; however, the Color Sampler Tool is capable of taking up to four samples per image (as demonstrated in fig. 1A). Red (Magenta) was not also used in the calibration procedure because a is a continuous green-red channel of the chromatic spectrum, where green = 0% and red (magenta) = 100% in the CIELAB system. Correction factors were incurred to achieve ‘pure green’, with L = 55, a = -38, and b = 31, and adjustments were made in Image > Adjustments > Brightness/Contrast and Color Balance for uncalibrated values of a only (fig. 1A). Adjustments were made to derive the colour balance standard for Brightness (Lightness in Photoshop, or L) and Color Balance (across Color Levels representative of a and b). The calibration was performed with adjustments in cycles until the correction factors were fully realised. Error was assessed based on a comparison of uncalibrated values to the calibration standard. Finally, subsequent to calibration, selecting Image > Histogram brings up information on the lightness and colour of images, which can be used to qualify colour (fig. 1B). Numbers are originally presented as values out of 255 and were subsequently converted to proportions.
Fig. 1 – Calibration method for a.
Fig. 1 – Méthode d’étalonnage pour a.
A: Color Sampler Tool and Color Balance. B: Histogram-based quantification.
A : Echantillonnage et balance des couleurs. B : Quantification basée sur l'histogramme.
9Even though a single digital camera was used throughout this study, there is a range of error in the uncalibrated images. Notably, sampling of No. 14 (green) revealed a total variance from a respective L/a/b of 55/-38/31 ‘pure’ colour standard for green that ranged from 10 (Site 1) to 39 (Site 9). Accordingly, the total of adjustments ranged from 19 to 88 at these sites, taking from 4 to 8 cycles for full calibration. Most images required 7 cycles of correction in the complete calibration procedure. Often, most correction was required for brightness (reduction), followed by a (increase), and last by b (reduction). Only in two cases were corrections not needed, namely for L at Site 1 and b at Site 13. In all cases, a had to be adjusted in the calibration procedure.
10When images were uncalibrated, it appeared that Site 9 was the greenest (Mean = 48.66%). However, with calibration, this changed and Site 10 became the greenest (Mean = 51.83%). Of the calibrated results, the greenest sites are Sites 10 and 3, whereas Sites 12, 8, and 11 are the least green (fig. 2A). In all cases, greening was reduced with calibration. Overall findings reveal varied levels of greening, even on north-facing walls. The standard deviation (Std Dev) is overall very small (< 4%). With calibration, the values of this measure increase at all sites (fig. 2B). Values are greatest for Sites 9, 7, 8, 11, and 12 (all located on north-facing walls), varying between 2.99% and 3.42%.
Fig. 2 – Results for calibrated and uncalibrated images.
Fig. 2 – Résultats pour les images calibrées et non calibrées.
A: % Mean. B: % Std Dev.
A : Moyenne (en %). B: Déviation standard (en %).
11The channel that is most variable and needs most correction is L, followed by a (and least for b). Previous studies also conveyed this result; for example, M. Thornbush (2008) found Mean calibrated values to be different from uncalibrated values, particularly in overcast conditions, as in this study. She also discovered that chromatic values were less altered with calibration, especially for b. It is necessary to calibrate images, as in the current study, since they would appear greener than they really are and possibly distort the magnitude of greening in quantitative (re)photography. M. Thornbush (2010a) revealed that using different cameras (such as Nikon Coolpix digital cameras) affects lightness more than chromatic results. It is not enough to simply take digital photographs and bring up histograms using software, such as Adobe Photoshop. The calibration procedure does affect the results and is an important consideration in this type of quantitative research.
12Localised differences in greening, even along the south side of the street, could be due to microclimate (even though care was taken to select north-facing sites as test sites), cleaning, or even the type of limestone used. For example, W.J. Arkell (1947, p. 160) noted that Pleurococcus has a predilection in particular to Taynton and Milton stones.
13Others have also found that climatic variables seem to govern the appearance particularly of algae on walls. For example, R. Rindi and M.D. Guiry (2002) portrayed the climate of western Ireland as having high rainfall and humidity, which helps to support both the richness and diversity of algae. They found the species Trentepohlia cf. umbrina usually concentrated on north-facing old limestone walls in the campus of NUI, Galway. Trentepohlia cf. umbrina did not vary significantly across months or seasons at a local spatial extent. These algae naturally grow very slowly, where snow and ice are generally absent and rainfall and humidity remain relatively high throughout the year. They found that moisture is particularly important in the field, and dry periods can limit algal reproduction. I. Grondona et al. (1997) observed a biopreservation effect due to constant humidity, where green alga appeared under crusts that acted to deter sandstone weathering. M. Flores et al. (1997) examined bacteria, yeast, and microalgae from samples on limestone surfaces in the old quarter of Alcalá de Henares in Madrid, Spain, where they are exposed to air pollution. These growths were especially evident in fissures and cavities in areas where evaporation is low, due to protection from wind and direct solar radiation, and moisture is retained.
14The results obtained in this study can only be used as exemplary due to a lack of complete control of microclimate. Even though an effort was made to select sites on the same street, with chosen test tests on the south side of the street and north-facing, it is still not possible to completely exclude variations in local temperature, rainfall, wind, and so on. Microclimate plays a complex role in bioweathering. First, moisture and sunlight control the growth of flora (including microflora). Wet and shaded locations are more conducive to algal growth, for example, whereas sunlit places can help establish lichens. Similar microclimates were compared in this study by controlling the orientation of test sites (north-facing) and the height at which photographs were taken (some 1 m above ground level) along the same street. Photography was limited to the morning hours of one season in order to control for the brightness of sunlight, and taken on a dry day.
15A ‘patchy’ distribution across rock surfaces increases Std Dev values and could be indicative of microbial and floral colonisation, such as the establishment of lichens and mosses, conveying a ‘spotted’ patchy pattern, which was evident in Oxford after the first year of a sensor exposure programme (Thornbush and Viles, 2004b). Moss growth, in particular, is implicated by higher Std Dev values in the current study. This is especially evident in rougher surfaces, where moss sprouts from depressions, such as pits, on surfaces. Calibrated images are better able to capture this pattern, as % Std Dev values increased with the calibration procedure (e.g., especially at Site 12 in fig. 3A and 11 in fig. 3B). This could be due to shadows on north-facing buildings (Thornbush, 2008). However, streaking could also contribute to this effect, either due to salt accumulations or associated with the rainwashing of soiled surfaces (see fig. 3B). M. Thornbush (2010a) also deciphered this pattern at the Ashmolean Museum of Art and Archaeology, University of Oxford before cleaning was performed and soiled surfaces portrayed uneven surface brightness due to rainwashing, as for example on columns. Greening by algae is also patchy in distribution (rather than a uniform biofilm cover), with accumulations appearing near water gutters and spouts in locations where there is greater water availability (fig. 3C). This does indicate the preference of algal growth in wet places, especially where shading provides the opportunity for a longer period of wetting. It is possible to use this method to quantity areal greening of buildings elsewhere with similar climate and stone types; however, caution should be taken for comparisons, particularly because of the effects of microclimate.
Fig. 3 – Images of sampled walls.
Fig. 3 – Images des murs échantillonnés.
A: Site 12; B: Site 11; C: Site 3.
A : Site 12 : B : Site 11 ; C : Site 3.
16Further research is required for a comparison of greening during drier periods, as in the summer, since a cross-seasonal comparison would be relevant to the biological colonisation of surfaces. Moreover, a long-term study may be of interest in order to decipher any environmental effects (especially of microclimate). For example, algal accumulation in wetter years; unfortunately, the local climatic record previously available from the Radcliffe Meteorological Station at Green College, Oxford does not extend beyond 2006. In terms of another biological study, it would be interesting to apply this technique to surface colonisation by (epilithic) lichens, since this method can quantify whiteness through L and yellow colouration through b. The influence of lichens is evident in the present study (fig. 1B and fig. 2B). This could affect values of lightness (through light colouration of some species of lichen) and colour dimensions (of green-hued lichens). Lichens also grow in circular patterns and could (as biological growths) contribute to the patchy (spotted) distribution addressed earlier. Finally, the O-IDIP method can further be applied to measure surface reddening through a, as with the measurement of surface accumulations of iron.
17The calibration procedure was attempted here this time using a ColorChecker, since this was not possible in a previous study (namely, Thornbush, 2008). This approach conveyed greater values for a, suggesting less greening of surfaces than uncalibrated digital photographs taken in an overcast condition during the winter. This method appears very effective in detecting the growth of moss within surface depressions on building ashlar through augmented values of Std Dev, particularly on north-facing walls. In this way, it is also quite effective for identifying rainwashed areas through the appearance of streaks (‘rainstreaks’), which develop as surfaces become differentially rainwashed, and are evident even in the a channel. There does not seem to be a significant effect of north-south aspect in this study, although more sites (particularly south-facing sites) are needed to further test this conclusion.
This paper was spurred by a request from Bruce Ford of the National Museum of Australia. The author is very grateful to S.E. Thornbush for fieldwork. Thanks also to two reviewers and Gilles Arnaud-Fassetta who edited the French texts.