Long-term surveillance of Phytophthora cinnamomi reveals no evidence of increased detections and new insights for monitoring and management

Study area

We used P. cinnamomi detection data from southern Sydney conservation reserves for this study (Table 1). The reserves were grouped geographically into (A) Dharawal Reserves; (B) Royal/Heathcote/Garawarra Reserves; and (C) Georges River/Kamay Botany Bay Reserves (Fig. 1e, Table 1). The study area is within the Sydney Basin and site geological characteristics relate to topographic elements: elevated plateaux support infertile sandy loams derived from Hawkesbury sandstones, Quaternary alluvium occurs in the headwaters of streams while unconsolidated sand or silt sediments are respectively associated with aeolian deposits or floodplain streams (Tozer et al. 2010). Vegetation structural categories were generalised as Sclerophyll Forest, Sclerophyll Woodland, Dry Heath, Scrub, Upland Swamp or Cleared land. The climate is temperate, and features warm summers and cool winters with mean annual rainfall of at least 1000 mm (http://www.bom.gov.au/climate/australia/cities).

Fig. 1.

Detection of Phytophthora cinnamomi in Dharawal Reserves in (a) 2008, (b) 2014, (c) 2022 (labelled A). Detection of P. cinnamomi in (d) 2006–2008 for Royal, Heathcote, Garawarra Reserves (labelled B) and Georges River and Kamay Botany Bay National Parks (labelled C). All southern Sydney conservation reserves are shown in (e) with an inset map showing the study location in Australia.

All of the conservation reserves are proximate to metropolitan Sydney. However, they experience different levels of human visitation. Estimated visitation to Royal National Park in 2008 was five-fold greater (3.6 million) than Kamay Botany Bay National Park (0.7 million). All other reserves in our study area were outside the top 20 visited conservation reserves (Anon 2019). We therefore categorised visitation as ‘high’ for Royal/Heathcote/Garawarra, ‘medium’ for Georges River/Kamay Botany Bay and ‘low’ for Dharawal Reserves.

Data sources and field collection

We compiled P. cinnamomi detection and field visual assessment data along with explanatory environmental variables from surveys undertaken in southern Sydney reserves during 2006–2008 (referred to as the ‘2008’ timepoint; n = 147; Burgess et al. (2017) and E. Liew unpubl. data). A new cohort of sites (n = 110) was sampled in a separate 2014 study focused on Dharawal Reserves (Craven and McDougall 2015). We subsequently resurveyed the 2014 sites (n = 110) between October 2021 and February 2022 (referred to as the ‘2022’ timepoint) to assess changes in disease incidence. Soil sampling followed Botanic Gardens of Sydney PlantClinic protocols (Suddaby and Liew 2008). When we resampled Dharawal Reserves in 2022, we navigated to site coordinates and attempted to sample beneath individuals identified in the 2014 survey. However, if the individual was dead or absent, we sampled to match the genus or family. We collected four to eight subsamples around the dripline of an identified plant individual. We removed leaf litter and sampled soil and root material to a depth of 10–15 cm. Soil collection was via a trowel in 2014 and via aluminium corers (2 cm diameter) in 2022. With both implements, sampling aimed to collect fresh root material to maximise P. cinnamomi detection (Newell 1998). Subsamples were bulked and mixed in a labelled clip seal bag. The bulked sample volume was approximated at ‘two cups of soil’ (Suddaby and Liew 2008). Samples were kept cool and delivered within 7 days (or 48 h in 2022) to the PlantClinic Laboratory at the Botanic Gardens of Sydney. Our field sampling followed strict hygiene protocols. Footwear and sampling equipment were treated with 70% ethanol (methylated spirits) after collection at each site to minimise contamination risk at subsequent sites. In 2022, we used a separate corer for each site with corers washed on a hot dishwasher setting and treated with 70% ethanol prior to reuse.

Visual site health assessments followed an ordinal scale at each site: healthy = 0–4% of all plants with symptoms of foliage senescence, branch or individual death; slight = 5–25% plants with symptoms; moderate = 26–50% plants with symptoms; severe = more than 50% plants with symptoms. Sites with slight to severe symptoms (>5%) were pooled and identified as displaying visual signs consistent with P. cinnamomi infection for our analyses. Sites were visually assessed at all sampling epochs and observers used the same visual assessment scale across surveys. The Dharawal Reserves sites sampled for P. cinnamomi presence in 2008 differed from those used in 2014 and 2022.

Laboratory techniques

Laboratory techniques used to detect P. cinnamomi at the Botanic Gardens of Sydney PlantClinic have evolved over the study period. The 2006–2008 and 2014 samples were processed by baiting samples with blue lupin (Lupinus angustifolius) seedlings (Pratt and Heather 1972) on Phytophthora-selective agar medium (PSM) followed by morphological identification by microscopy (Suddaby 2008). The 2021–2022 sampling period spanned a changeover in laboratory procedures at the PlantClinic: 32 samples were processed using a community DNA extraction approach from samples baited with blue lupin seedlings followed by Phytophthora specific Polymerase Chain Reaction (PCR) of the Ypt1 gene region (Schena et al. 2008). A total of 78 samples were processed using PARP 10 agar baits (Nakova 2010) that were modified by using clarified V8 as the basal medium. Detection relied on PCR using a Phytophthora genus specific Taqman™ assay (Laurence et al. 2023). We did not find any documentation comparing detection sensitivities for morphological and PCR identification at the PlantClinic.

Data analysis

Phytophthora cinnamomi detection over time

Sites within the Dharawal Reserves that were subject to repeat surveys transitioned in detection status over time (Fig. 1a–c). In 2014, P. cinnamomi presence was detected at 50 of 110 sites (0.45 as a proportion) with a further five sites positive for the Phytophthora genus. In 2022, P. cinnamomi presence was detected at 51 of the same 110 sites (0.46 as a proportion) with a further site positive for the Phytophthora genus. Overall detection appeared stable (χ22=2.2, P = 0.330; Fig. 2), however a positive status at individual sites changed over time reflecting either the passage of a disease front or variable detection. In 2008, a separate cohort of sites in Dharawal Reserves were positive for P. cinnamomi at 6 of 26 locations (0.23 as a proportion).

Fig. 2.

Estimated probability of Phytophthora cinnamomi presence (±95% CI) over time in Dharawal Reserves averaged over landform type. Presence probabilities for Georges River/Kamay Botany Bay National Parks and Royal/Heathcote/Garawarra Reserves are available for the first time point only.

We assumed laboratory detection methods were comparable across the longitudinal study, as has been found in a related formal comparative study (Sena et al. 2018a). Both raw and estimated means notably appeared to increase between 2008 and 2022 but the large confidence intervals in Fig. 2 indicate considerable trend uncertainty.

Inconsistent pathogen detection and the role of sentinel species

Across 2008, 2014 and 2022, we found discrepancies between field observations of site-level P. cinnamomi symptoms and detection via laboratory analyses. Agreement between detection methods was 38% when laboratory analyses were negative and 57% when laboratory analyses were positive for P. cinnamomi. In exploratory analyses, we found that performance of field visual assessment was poor (AUC = 0.46 (95% CI: 0.41–0.52)). When we focused only on sites where Xanthorrhoea sp. was identified as a host species, performance of visual assessment improved (AUC = 0.61 (95% CI: 0.47–0.75)). Agreement between detection methods increased to 52% when laboratory analyses were negative and 64% when laboratory analyses were positive for P. cinnamomi.

Environmental variables as predictors of P. cinnamomi detection

Based on our primary model that contained a spatially structured random effect, we found no evidence of a relationship between P. cinnamomi detection and landform, vegetation type, distance to waterway, distance to roadway or visitation (based on conservation reserve clusters) (P > 0.10). We also fitted a model without a spatially structured random effect as a sensitivity analysis that explored spatial confounding. This model had larger AIC than the primary model. The sensitivity analysis yielded evidence of a relationship between distance from waterway and detection probability: the odds of P. cinnamomi detection increased by 19% (95% CI: 2–39%) for every 100 m towards a waterway (χ12=6.7; P = 0.010). Confounding between spatially structured covariates and the spatial random effect may obscure the effect of covariates (Hodges and Reich 2010) and we suspect this has occurred in our primary analysis. Distance to waterway was the only effect that differed qualitatively between models.

The primary model did not provide evidence of human visitation effects across the reserve categories (χ22=1.8, P = 0.416). The mean estimated probability of P. cinnamomi detection in Dharawal Reserves (with the lowest visitation) was 0.21 (±95% CI: 0.06–0.54) in 2008, 0.35 (±95% CI: 0.16–0.60) in 2014 and 0.50 (±95% CI: 0.26–0.74) in 2022. These estimates were averaged across landform type. In 2008 the infected proportions were 0.18 (±95% CI: 0.06–0.46) for the medium visitation reserves (Georges River/Kamay Botany Bay National Parks) and 0.41 (±95% CI: 0.19–0.67) for the highest visitation reserves (Royal/Heathcote/Garawarra Reserves) (Figs 1d and 2). A trend of positive detection prevalence and Reserve visitation may be obscured by large confidence intervals.

Estimating future sampling requirements to improve future monitoring

Given the uncertainty of trend estimates in the initial monitoring, we explored requirements of future monitoring to detect changes in P. cinnamomi presence, should these occur. Using the raw detection proportion of 0.46 in 2022, we found that a repeat of 110 samples at the next timepoint would only be able to detect an increase in pathogen presence of ≥47%. More intense sampling of size n = 300 — in each of two future sampling periods — would be more sensitive to increasing P. cinnamomi levels and could detect when pathogen incidence increased by ≥29%.

Phytophthora cinnamomi detection over time

Infection by P. cinnamomi represents an ongoing threat within southern Sydney conservation reserves. There is uncertainty as to whether the threat is intensifying. The estimated marginal means for the proportion of P. cinnamomi positive sites in Dharawal Reserves suggest that prevalence may be increasing over time. However, this trend was statistically undetected and characterised by large confidence intervals likely reflecting limited sample size. Detection appeared spatially and temporally heterogeneous without evidence of clear disease fronts. Such patterns conform with other observations for NSW (McDougall and Summerell 2003). We did not find evidence of increasing P. cinnamomi incidence despite greater fluctuations in climatic moisture prior to the 2022 survey (e.g. below average annual rainfall in 2019 and above average annual rainfall in 2021), recent prescribed fire and wildfire activity (during the 2001–2002; 2005–2006; 2015–2016; 2018–2019; 2020–2021 fire seasons), and ongoing human visitation and feral animal activity.

Baseline monitoring of disease threat status is imperative for local, state and national conservation management. Reserve-level monitoring directs local suppression effort (e.g. Wilson et al. 2020) and contributes to state and national threat abatement planning and monitoring. A review of the Commonwealth Environmental Protection and Biodiversity Conservation Act 1999 found inadequate monitoring of key threatening processes (Samuel 2020) and therefore an incapacity to demonstrate protection of threatened species and communities. Ongoing monitoring of pathogen detections and consideration of spatial patterns in multiple reserves across the potential disease distribution would directly address the national threat status. However, our study has demonstrated that substantial within-reserve sampling effort is required to confidently observe changes in P. cinnamomi prevalence. Furthermore, sites must be revisited at ecologically relevant intervals: Botanic Gardens of Sydney protocols recommend revisits every 2 years or after high-risk disturbance such as fire or flood (Suddaby and Liew 2008). In addition, monitoring activities should be timed to maximise detection capacity. For example, teams should mobilize after wet periods — when zoospore activity is more likely — and avoid soil collection during drought conditions (Drenth and Sendall 2001).

Governance of monitoring programs must include evaluation, adaptation and calibration of methodological advancements to improve reporting outcomes (Lindenmayer et al. 2012). In our case, formal comparison of P. cinnamomi detection sensitivity when laboratory assay methodologies change will aid future interpretation of pathogen trend analysis.

Importantly, a P. cinnamomi threat status that is not clearly intensifying does not imply negligible or minimal ecosystem effects. Selective and cumulative elimination or damage of susceptible plant species has implications for biodiversity value, habitat provision and ecosystem function (Weste et al. 2002; Cahill et al. 2008). In addition, disease expression in vegetation communities may be compounded by multiple disturbances such as prescribed fire or wildfire (Metz et al. 2013; Moore et al. 2014). Interactions between disease incidence and climate change may disproportionately affect symptomless or moderately susceptible host species if climatic conditions become more favourable for P. cinnamomi (Thompson et al. 2014; Burgess et al. 2017).

Inconsistent pathogen detection and the role of sentinel species

Monitoring for P. cinnamomi disease generally involves both visual assessment of susceptible individuals for dieback symptoms and laboratory testing of soil samples (Summerell and Liew 2020). However, we found low congruence between field visual assessment and laboratory soil testing. Inconsistencies between detection methods may be explained by combinations of several factors. Firstly, host susceptibility or disease expression may be variable: the host may be infected by P. cinnamomi but appear asymptomatic (e.g. McDougall and Summerell 2003; Wan et al. 2019). While a suite of species has been identified as susceptible to dieback expression following P. cinnamomi infection (Cahill et al. 2008), researchers have increasingly reported susceptibility spectra within families, genera and species (Shearer et al. 2013). Alternatively, low congruence between field and laboratory findings may be explained by erroneous attribution of dieback to P. cinnamomi disease expression when water stress or other stressors may be the cause of symptoms (Jurskis 2005; Anon 2018). And where symptoms appear severe, P. cinnamomi infection may be the cause, however the pathogen may have migrated elsewhere and evade detection in soil samples (Weste and Marks 1987). Furthermore, P. cinnamomi is difficult to isolate from soil samples and may evade detection due to insufficient sampling (Pryce et al. 2002). Finally, laboratory detection may be imperfect with the testing process returning false negative (low specificity) or false positive (low sensitivity) results (Davison and Tay 2005; O’Brien et al. 2009).

While direct sampling of soil or plant material remains a centerpiece of monitoring P. cinnamomi incidence, this is expensive and difficult (Newby et al. 2019) and detection is uncertain under conditions of patchy disease expression. Indirect detection methods using remote sensing (Landsat satellite and orthorectified aerial imagery) have tracked temporal trends in Phytophthora dieback (Cardillo et al. 2018). However, low congruence between laboratory and on-ground visual assessment in our study suggests that aerial interpretation would be unreliable as a rapid detection approach for our coastal NSW study area. Multispectral and hyperspectral imagery have also successfully identified symptomatic vegetation. However identification of asymptomatic or marginally affected vegetation and separation of water stress and disease expression have proved challenging (Newby et al. 2019).

Sentinel species may differ across regions: in Sydney, members of the genus Xanthorrhoea may be candidates (Keith et al. 2012; Summerell and Liew 2020). These are potentially useful sentinels because they are long lived, conspicuous and susceptible. We found marginally greater alignment between pathogen detection and disease expression when we opportunistically analysed a subset of sites where Xanthorrhoea were identified as potential hosts (observers frequently did not distinguish among species within the genus). Xanthorrhoea hosts were often among several identified hosts and field observers did not attribute health response to single hosts. At present, we conclude that visual assessment is insufficiently informative to justify this as a sole monitoring approach. However, further research that explicitly links individual hosts and field health observations may clarify the efficacy of sentinel species in site selection for disease monitoring. We advocate a dual surveillance strategy for P. cinnamomi that explicitly measures the field condition of sentinel species to estimate disease expression and undertakes laboratory assay of the soil pathogen to identify disease potential. This dual approach will produce complementary insights into the dynamic status of the disease in the landscape and inform disease management.

Environmental variables as predictors of P. cinnamomi incidence

We found evidence that detection was higher proximate to waterways (in the non-spatial model) but was unrelated to road and track formation, vegetation type or landscape position. We postulate that, in the absence of active suppression, P. cinnamomi incidence may continue to migrate upslope from creek and drainage lines in southern Sydney conservation reserves. Other studies have demonstrated that waterways facilitate pathogen transport (von Broembsen 1984; Smith et al. 2009; Hüberli et al. 2013) and disease transmission may subsequently occur via direct root contact in the soil matrix (Hill et al. 1994). In light of our findings, water used for firefighting and road-building within uninfected catchments should be sourced from within the catchment (Hüberli et al. 2013) or from uninfected water sources (e.g. chlorinated mains water (Anon 2022b)). Additionally, creation of new vehicular and walking tracks should be avoided and unnecessary extant tracks should be closed where the pathogen is believed to be absent (e.g. Wilson et al. 2000).

Comparison of raw disease detection among low, medium and high visitation reserves (respectively Dharawal Reserves; Georges River/Kamay Botany Bay Reserves and Royal/Heathcote/Garawarra Reserves) suggests that disease incidence may increase with increasing human access. Consequently, areas considered disease-free should be subject to catchment-scale quarantine and hygiene measures.

A management strategy for Dharawal Reserves

Our study has found that P. cinnamomi is widely distributed in southern Sydney conservation reserves and that new detections within Dharawal Reserves have occurred within the monitoring period (2008–2022). Long-term studies have also indicated that P. cinnamomi infection causes elevated mortality rates among susceptible plant species (e.g. Weste 2003). However, there is considerable uncertainty about change in disease incidence and detection confidence at the site scale. This uncertainty makes management of the disease a challenge in eastern Australia where disease expression is typically diffuse (McDougall and Summerell 2003). Due to widespread and long-term detection of P. cinnamomi in southern Sydney and the limitations of current control methods (Sena et al. 2018b), we surmise that invasion has progressed beyond prevention, eradication and containment to the asset protection phase for management actions (Fleming et al. 2017). Nonetheless, ongoing monitoring is essential to firstly inform future management and containment of P. cinnamomi as a key threatening process and secondly protect natural assets. The outcomes of this study suggest how managers may improve confidence in threat status monitoring and provide indicative costs. A sample size analysis indicated that managers would need to increase the sample size from 2022 (n = 110) nearly three-fold (n = 300) — and over two time periods — to detect a 29% increase in Reserve-level P. cinnamomi presence. Such an increase in sampling effort has considerable resourcing implications. In 2023, the Botanic Gardens of Sydney PlantClinic laboratory costs were AUD175 per sample (https://www.botanicgardens.org.au/our-science/our-services/plantclinic-plant-disease-diagnostics). In addition, a two-person team budgeted nine field days in 2014 and 11 field days in 2022 to collect 110 samples across Dharawal Reserves.

The study has identified several pertinent management issues at the reserve level. A monitoring program that resamples sites on a 2–5 year rotation would detect changes in threat status more effectively than sampling at the 6–8 year interval of this study. In any case, escalating P. cinnamomi detection should trigger a reassessment of the prevailing management approach. Where the pathogen has been detected recently, managers should monitor the impacts and protect natural assets. Chemical control of P. cinnamomi, using phosphorous acid (phosphite), has been effective in mitigating population- and community-level impacts in Western Australia (Barrett and Rathbone 2018; Keighery et al. 2023). Phosphite can be sprayed or injected and induces defense responses in infected plants (Garkaklis et al. 2004). Barrett and Rathbone (2018) found that species assemblages and vegetation structure in phosphite-treated infested communities were more similar to healthy than infested non-treated communities. Keighery et al. (2023) found that Banksia anatona survival after fire was greater when seedlings were recurrently sprayed with phosphite compared with untreated seedlings. While phosphite application may improve ecosystem function in infected areas, its effectiveness under NSW climatic conditions and the optimal timing of application require further research. There may also be cumulative collateral impacts from repeated addition of phosphite to low-fertility soils such as Hawkesbury sandstone derived soils of the Sydney region (Lambers et al. 2013).

We suggest that control methods in disease-affected high-priority conservation areas could focus on discrete populations of slow-turnover susceptible species such as Xanthorrhoea resinosa. Xanthorrhoea species are ‘old growth’ components of forest, woodland and heath communities in Sydney conservation reserves (Regan et al. 2011; Wilson et al. 2020). Xanthorrhoea species may also be considered umbrella species (Noss 1990), providing habitat (e.g. Wilson et al. 2020) and forage — especially in the immediate post-fire environment (e.g. Brown and York 2017) — and ameliorating conditions (e.g. Petit and Dickson 2005) for many co-occurring species. The susceptibility of Xanthorrhoea species to P. cinnamomi damage and their conservation multiplier effects, qualify these as focal species (refer to Lambeck (1997) for a related discussion) for pathogen control and monitoring. Given the difficulty and expense of land-extensive pathogen detection and management, a monitoring and control approach that focuses on sentinel and umbrella species may be appropriate. Furthermore, phosphite trials (with careful observations of off-target damage) in discrete areas could prioritise Xanthorrhoea populations where P. cinnamomi is an active threat (Wilson et al. 2024).