Over the past 5 years, a number of cases of legionellosis in Scotland have been associated with compost use; however, studies investigating sources of infection other than water systems remain limited. This study delivers the first comprehensive survey of composts commonly available in the UK for the presence of Legionella species. Twenty-two store-bought composts, one green-waste compost and one home-made compost were tested for Legionella by culture methods on BCYE-α medium, and the findings were confirmed by macrophage infectivity potentiator (mip) speciation. Twenty-two of the samples were retested after an enrichment period of 8 weeks. In total, 15 of 24 composts tested positive for Legionella species, a higher level of contamination than previously seen in Europe. Two isolates of Legionella pneumophila were identified, and Legionella longbeachae serogroup 1 was found to be one of the most commonly isolated species. L. longbeachae infection would not be detected by routine Legionella urinary antigen assay, so such testing should not be used as the sole diagnostic technique in atypical pneumonia cases, particularly where there is an association with compost use. The occurrence of Legionella in over half of the samples tested indicates that compost could pose a public health risk. The addition of general hygiene warnings to compost packages may be beneficial in protecting public health.
Legionella pneumophila was first established as a human pathogen after an outbreak of pneumonia at a meeting of the American Legion in Philadelphia in 1976, which led to the hospitalization of >100 ex-servicemen and, subsequently, 29 fatalities. To date, >50 species of Legionella have been identified; almost half have been cited as causative agents of human disease . Infection with Legionella may be asymptomatic, or present as legionellosis in one of two distinct clinical manifestations: Pontiac fever—a self-limiting influenza-like illness; or Legionnaires’ disease (LD)—a more serious pneumonia that can be fatal.
Most cases of infection reported throughout Europe are caused by L. pneumophila, which was responsible for 96.5% of culture-confirmed cases of LD reported to the European Legionnaires’ Disease Surveillance Network (ELDSNet) in 2010 . The incidence of infection with Legionella longbeachae in Europe has historically been low; since 1984, L. longbeachae has been cited as the causative agent in only 11 cases of infection in the UK, seven of these occurring in Scotland . However, four of these cases (36.4%) occurred in Scotland between 2008 and 2010, and in all cases the patient had been in contact with commercially available compost before the onset of symptoms .
Cases of LD where L. longbeachae is the aetiological agent are not limited to the UK. The incidence of human infection with L. longbeachae is much higher in the southern hemisphere than in Europe; rates of infection are more equivalent to those for L. pneumophila. A review of legionellosis survey data in Australia from 1996 to 2000 reported that 42% of cases were attributable to L. longbeachae, as compared with 51% cases where the causative agent was L. pneumophila . In New Zealand, the Ministry of Health found that, in 2011, L. longbeachae was responsible for more cases than L. pneumophila, with 42% and 30% of laboratory-reported cases of infection, respectively . Human infection with L. longbeachae has also been noted in Thailand, where Phares et al.  found that L. longbeachae was responsible for 5% of clinically defined cases of pneumonia in a rural district, whereas L. pneumophila was not reported.
Steele et al.  isolated L. longbeachae serogroup (Sg) 1 from potting mixes in South Australia in the late 1980s, and later detected L. longbeachae and other species of Legionella in almost three-quarters of the investigated potting soils manufactured in Australia . Cramp et al.  were the first to identify L. longbeachae Sg 2 as the causative agent in an outbreak of Pontiac fever that was traced to exposure to an aerosolized potting mix in a horticultural nursery where the patients had worked.
In addition to studies in Australia [8, 10] and New Zealand , pathogenic and non-pathogenic Legionella species have been isolated from potting soils in Japan , Switzerland , and Greece . Of all species isolated, L. pneumophila, L. longbeachae, Legionella bozemanii, Legionella micdadei and Legionella anisa are known to cause disease in humans ; however, with the exception of one case of L. pneumophila Sg 1 , only L. longbeachae has been directly linked to cases of infection where compost or potting soils were cited as the source of infection [7, 9]. Although Legionella species have been isolated from UK composts [3, 15], studies have only looked at very small sample sizes, and to date, a study examining the prevalence of Legionella in a wide range of UK composts has not been undertaken. The aim of this study was therefore to address this knowledge gap and determine the incidence of Legionella species in commonly available UK composts.
Twenty-two branded composts, purchased from retailers in the UK, were analysed for the presence of Legionella species. Common ingredients of the tested brands included a mixture of sphagnum moss peat, composted organic material, sand, vermiculite, and loam. Six of the composts were designated as peat-free. In addition to the branded composts tested, a sample of green-waste compost treated to PAS 100 specifications  and one sample of home-made compost, matured from garden waste, ash, rainwater, and limited household waste, were tested.
To ensure that a representative sample was used, compost was taken from five areas of each bag, in cores 4 cm in diameter and 6 cm in length. This material was then mixed well in a sterile container prior to analysis. Each compost sample was tested three times, and the results were pooled to enhance detection rates. The test procedure utilized was adapted from those previously published [3, 7, 12]. Briefly, 5 g of compost was added to 50 mL of sterile dH2O before being shaken for 1 h at 150 r.p.m. After 15 min of settling, a 200-μL aliquot was taken and acid-treated with an equal volume of 0.2 M HCl/KCl (pH 2.2) for 15 min. Acid treatment was carried out in an attempt to suppress unwanted compost flora; legionellae are relatively acid-resistant. The treatment time of 15 min was deemed to be appropriate after initial testing revealed high levels of bacteria and fungi in the compost samples. A ten-fold dilution of the acid-treated sample was prepared in sterile dH2O, and 50 μL of this was plated onto Buffered Charcoal Yeast Extract (BCYE) Agar, supplemented with Legionella BCYE-α growth supplement (potassium hydroxide buffer, ferric pyrophosphate, l-cysteine-HCl, and α-ketoglutarate; SR0110; Oxoid, Basingstoke, UK) and GVPC (glycine, vancomycin hydrochloride, polymyxin B sulphate, cycloheximide; SR0152; Oxoid). The remainder of each sample was left at 30°C, as described by Koide et al. , to allow ‘enrichment’ of Legionella by any amoebae present in the sample. After 8 weeks, these samples were reprocessed with the methods described above.
BCYE plates were incubated for 3–7 days in humid conditions at 37°C, and examined regularly for potential Legionella colonies by use of a light microscope (Olympus CHB, Southend-on-Sea, UK) with a cold light source (Schott KL1500; Schott UK Ltd, Stafford, UK); Legionella colonies have a distinctive ground-glass appearance. Presumptive colonies were subcultured onto fresh BCYE-α containing GVPC, and cysteine-negative agar (BCYE-α supplement without l-cysteine; SR0175; Oxoid), as a negative control, as Legionella species require cysteine to grow. All colonies that grew on BCYE-α but not on cysteine-negative agar were subcultured to ensure purity, and stored on Microbank beads at −80°C (Pro-lab Diagnostics, Wirral, UK). After all compost samples had been tested, potential Legionella colonies were analysed by use of PCR with Legionella-specific macrophage infectivity potentiator (mip) primers . Successful PCR products were determined by gel electrophoresis with a 1% agarose gel, and purified by poly(ethylene glycol) precipitation. Sequences were determined with the LightRun sequencing service (GATC Biotech, Konstanz, Germany), and were run through the Public Health England (formerly the Health Protection Agency) Legionella mip gene sequence database (www.hpa.org.uk) in order to identify samples at the species level.
Serotyping of strains was carried out at the Scottish Haemophilus, Legionella, Meningococcus and Pneumococcus Reference Laboratory, Stobhill Hospital, Glasgow. Latex agglutination for L. pneumophila strains was carried out with the Legionella Latex Test for L. pneumophila Sg 1, L. pneumophila Sg 2–14, and Legionella species (Oxoid). Immunofluorescent antibody testing was then performed on strains with more than one serogroup by the use of antisera raised in guinea pigs and rabbits and fluorescently labelled anti-guinea pig and anti-rabbit sera.
Twenty-four compost types were tested for the presence of Legionella species by culture on BCYE-α and confirmation with PCR of the mip gene. As can be seen in Table 1, 15 of the 24 (62.5%) samples tested contained Legionella species. Ten of 24 (41.7%) were positive in the first round of testing, and, of the 22 samples that were retested after enrichment at 30°C, 13 (59.1%) tested positive after the 8-week incubation period. Owing to sample dilution, the detection limit of one colony per plate corresponded to 4000 CFU in 1 g of compost material; although this is quite high, it was deemed necessary to prevent inhibition of growth by other soil organisms. The maximum concentration of Legionella found was 4 × 104 CFU/g.
Table 1. Number (CFU/g) of Legionella species in compost samples tested directly or after 8–10 weeks of enrichment at 30°C
Initial culture (CFU/g)
Culture retested after 8–10 weeks (CFU/g)
GW, green waste; HM, home-made; NT, not tested; PC, peat-containing; PF, peat-free; Sg, serogroup.
Low percentage identity match (<98%) in the Health Protection Agency macrophage infectivity potentiator (mip) database.
Twelve species in total were identified in the compost samples according to mip gene speciation. These are shown in Table 1. The most commonly isolated Legionella species was Legionella sainthelensi, which was present in five of the 24 (20.8%) samples. This species was only isolated after the 8-week enrichment period, as can be seen in Fig. 1. Two isolates of L. pneumophila were found in the same compost sample; both tested negative with the latex agglutination kit (Oxoid Ltd, Basingstoke, UK). The strains were identified as L. pneumophila Sg 4 Portland and L. pneumophila Sg 1 OLDA by monoclonal immunofluorescent antibody testing. L. longbeachae Sg 1 was isolated from four of the compost samples, making it the second most commonly isolated named organism (Fig. 1). Three unnamed species were also isolated, Legionella 99-113, Legionella species K, and Legionella species A.
In total, there were eight isolates with low (<98%) percentage identity matches on the Health Protection Agency mip database. Comparison of the nucleotide sequences for these eight isolates revealed five potentially new organisms. According to BLAST: samples 9 and 16, Legionella 99-113, had a 100% identity match; sample 22 (initial and after enrichment), Legionella feelei, had a 99% identity match; and samples 10 and 12, Legionella gormanii, had a 99% identity match. Further work to establish the identity of these organisms is ongoing. Of the eight low-percentage match isolates, four were isolated from peat-free composts, and four from composts containing peat, as can be seen in Table 1.
Legionella species were detected at levels ranging from 4.0 × 103 to 4.0 × 104 CFU/g in 62.5% of the composts tested, suggesting that these organisms are common contaminants of UK composts. This contamination rate is higher than that found in compost surveys conducted in Greece  and Switzerland , where Legionella species were isolated from 27.3% (6/22) and 45.7% (21/46) of compost samples, respectively. These studies challenge earlier work by Steele et al. , who found Legionella species in 73% (33/45) of Australian composts tested, but failed to isolate the organism from European composts (14 from the UK, four from Greece, and one from Switzerland). Steele et al.  also tested compost components for Legionella species, and found the bacteria in 80% of composted bark and sawdust samples, but not in pure peat samples . Bark and sawdust are the main components of Australian compost, whereas peat has historically been a major component in European composts . In Japan  and Greece , all compost samples composed purely of peat were also negative for Legionella species, but samples containing peat mixed with other components have been found to contain these bacteria . The current study isolated Legionella species from two-thirds (12/18) of composts containing peat mixed with other components. These results suggest that, although peat alone may not support the survival of Legionella species, its addition to potting mixes does not prevent contamination with legionellae.
Traditionally, peat has been the dominant growth medium used in the UK and throughout Europe; however, it is likely that the composting market will increasingly move towards the use of peat-free composts as plans to eliminate the unnecessary use of peat by the year 2030 are implemented . Increased composting of green waste resulting from reduced landfill waste disposal, as required by the EU Landfill Directive 1999/31/EC, will provide alternative compost constituents. Standards are in place to ensure the quality of compost from green waste; however, there is no provision for Legionella species in current microbial standards , despite their isolation from composts and composting facilities in Australia  and Switzerland ; the latter study established green-waste collection and composting sites in Switzerland as an important reservoir for Legionella species. Lindsay et al.  hypothesized that the increased use of green waste and the decreased use of peat in commercial multipurpose composts in the UK, in relation to environmental concerns, may explain in part the increased incidence of L. longbeachae infection in Scotland.
The incidence of Legionella infection does not seem to be comparable with the presence of known pathogenic Legionella species in composts; for example, almost 17% of composts in this study tested positive for L. longbeachae, the causative agent in only 11 cases of infection in the UK since 1984 . This discrepancy may be attributable to cases of pneumonia caused by species other than L. pneumophila going undetected. Of the 12 different species of Legionella isolated in this study, at least eight are known to cause human disease. However, in Europe, the main diagnostic test used for the detection of LD is the urinary antigen assay, used in 81.9% of cases , which identifies only L. pneumophila Sg 1 with any degree of sensitivity . This issue of under-reporting of cases of legionellosis caused by species other than L. pneumophila was highlighted by Whiley and Bentham  and Lindsay et al. . Whiley and Bentham  also commented that individuals with Pontiac fever do not generally require hospitalization, and therefore Legionella infection, potentially related to compost, would not be diagnosed. In addition, the identification of currently unnamed Legionella species in this study reflects the diversity of strains in compost. The eight isolates with <98% mip speciation may represent new species and a new reservoir for infection.
It is important to note the relatively high limit of detection resulting from the methods used in this study. The high level of dilution and extended acid treatment were deemed to be necessary because of the high levels of bacteria and fungi present in the samples. Although different strains of bacteria will react differently to the acid exposure time, the isolation of 12 species from 62.5% of samples indicates that there is a diverse population of Legionella species present in UK composts. Although higher numbers of legionellae would perhaps have been obtained with a shorter treatment time, detection may have been negated by competing compost microflora. It is likely that Legionella species have only been identified in the most contaminated compost samples tested, and therefore a negative result does not rule out the presence of Legionella species at levels <4000 CFU/g. Low levels of legionellae in composts may help to explain the low incidence of human infection as compared with the relatively high contamination rate. Lindsay et al. isolated L. longbeachae from potting compost used by patients with L. longbeachae infection that had been stored in greenhouse conditions. The authors also noted that, in a preliminary study, compost stored in greenhouse conditions had increased levels of legionellae .
To an extent, the 8–10-week enrichment period in this study demonstrated the potential that increased temperature and humidity, as would be found in a greenhouse, have to enhance legionellae numbers. This enrichment technique was first used with compost samples by Koide et al. , based on previous work in which water samples containing amoebae were incubated to improve legionellae detection rates . Koide et al. directly isolated legionellae from ten of 24 samples taken from composted wood products and potting mixes. Initial samples were suspended in sterile water and incubated at 33°C for 2–3 months, allowing time for any amoebae and intracellular legionellae present to replicate to a level above the limit of detection. After the enrichment process, 22 of the 24 samples were positive for legionellae . In the current study, ten of the 24 compost samples were positive upon direct sampling; this increased to 13 of 24 after enrichment. However, as seen in Fig. 1, legionellae were not detected after enrichment in a number of samples that were positive upon direct plating, indicating that the technique was not always successful. A similar effect was seen by Koide et al. , whereby, upon direct plating, L. bozemanii was isolated from two samples of composted wood products, and Legionella birminghamensis was found in a potting mix sample; however, neither was detected after enrichment. In both studies, samples were not inoculated with amoebae; therefore, it is possible that suitable protozoa were not present in all of the samples, or in sufficient numbers to allow the successful replication of Legionella species. In the current work, samples 3, 9 and 15 showed the greatest increases in numbers of legionellae after the enrichment period; these three samples all contained L. sainthelensi, which was only isolated after the enrichment period. The results suggest that amoebal enrichment may be species-dependent. It is possible that the presence of amoebae in compost allows some Legionella species to multiply to potentially infectious numbers under the right conditions; however, a full-scale study into the effect of greenhouse conditions is needed to investigate this theory further.
The current study has shown that UK composts are commonly contaminated with Legionella species, many of which have been shown to cause human disease. Cases of non-L. pneumophila infection, which are generally on the increase globally , have been associated with compost use, primarily where the causative agent is L. longbeachae. O'Connor et al.  highlighted the fact that the presence of Legionella species in compost does not necessarily indicate that those handling it will become infected. During a case–control study examining the association of L. longbeachae infection with handling potting mixes, the authors used multivariate analysis to demonstrate that awareness of a potential health risk in using compost was protective against infection, although the authors could not demonstrate the effect that this knowledge may have on gardeners. The authors also found that not washing hands after gardening, before eating or drinking, was a risk factor for infection. Australia already has general hygiene warning labels on compost to educate users of potential risks. The work of O'Connor et al.  would suggest that this is a beneficial approach to protecting public health. It would therefore seem worthwhile considering a similar warning scheme in the UK and other countries, in order to increase public awareness. Such a scheme should highlight the need for good hygiene practices, including hand-washing after use and before eating, and a recommendation to open bags in a well-ventilated area.
In relation to cases of infection, clinicians should be aware of the increased incidence of cases of non-L. pneumophila infection, and, to allow for accurate diagnosis, urinary antigen testing should not be the sole diagnostic tool used in cases of community-acquired pneumonia, particularly if an association with gardening has been identified.