Forests increasingly suffer from pests and pathogens that negatively affect tree growth and productivity (Linnakoski et al., 2019). In Sweden, forests may be owned by private households, companies, or the state, but a degree of homogeneity in forest management is maintained through strict supervision by the national Swedish Forest Agency. Coniferous trees are prioritized for production in northern Sweden. However, the native Scots pine (Pinus silvestris) is damaged by twisting rust (Melampsora pinitorqua), which uses the European aspen (Populus tremula) as an obligate alternate host. Aspen is a highly regenerative and fast-growing early succession deciduous tree (Kassfeldt, 2009) that is rich in phenolic defense compounds including condensed tannins (CTs), which are considered mediators of stress tolerance (Ullah et al., 2017; Harding, 2019). Condensed tannins are bioactive phenolic polymers that vary in concentration with both biotic and abiotic factors including season, exposure, and host age (Rubert-Nason and Lindroth, 2021). Moreover, recent studies point at fast responses in tannin concentrations in response to fungal infestation (Ullah et al., 2019; Chowdhury et al., 2023). Leaf CTs thus vary considerably within a genotype, but within populations, high and low CT-producers may be distinguished (e.g. Decker et al., 2017; Bandau et al., 2021).

Melampsora rusts are biotrophic pathogens that alter physiological and metabolic processes in the hosts. Consequently, rust infection stunts aspen growth and reduces the value of pine timber (Tabor et al., 2000; May-De Mio and Ruaro, 2008; Toome et al., 2009; Cortizo, 2014; Gortari et al., 2018a; Gortari et al., 2018b). Rust fungi produce bright orange urediniospores that cause seasonal rust outbreaks in natural aspen stands and poplar plantations throughout Europe (Barrès et al., 2012). The risk of rust infection in pines has been linked to proximity to aspen stands, genotype, and soil fertility (Mattila, 2005). Therefore, during the 1960s and 1970s, Swedish forests were subjected to aerial treatments with herbicides to eradicate the less valuable aspen (Eidmann and Klingström, 1976). In addition, to increase pine growth and productivity, during the second half of the 20^th century, Contorta pine (Pinus contorta) was widely introduced in northern Sweden from Canada, where it is host to the local rust M. medusae (Elfving et al., 2001).

Melampsora pathogens have five sporulation stages in their life cycle tied to two alternate hosts (Newcombe and Chastagner, 1993; Mattila et al., 2001; Pei et al., 2005; Feau et al., 2009; Hacquard et al., 2013; Li et al., 2016). In temperate and boreal forests, a salicaceous host (poplar or willow depending on the rust species) becomes infested with dikaryotic (n+n) aeciospores. Throughout the growth season, the aeciospores develop and vigorously multiply on the leaves, initially as orange urediniospores (n+n) that subsequently become melanized “early (n+n)” and then (on senesced leaves) “late (2n)” teliospores. The teliospores are rich in lipids and glycogen, which enables the pathogen to tolerate winter frost (Feau et al., 2009). After undergoing karyogamy and meiosis, haploid basidiospores (n) are formed, which can then infect the alternate host (which may be a coniferous tree). Spermatogonia containing spermatia spores are then formed on the alternate host. The spores finally cross-fertilize to form dikaryotic (n+n) aeciospores that can infest another Salicaceae host (Hacquard et al., 2013).

Biomonitoring is essential for identifying and implementing appropriate management measures to control the build-up and spread of native and foreign pathogens (Choi and Park, 2019). Landscape-level information on abiotic and biotic damage to forests is therefore collected across Sweden by monitoring randomly selected individual trees and is made accessible through the Swedish National Forest Damage Inventory (SNFDI) database (Wulff et al., 2006). The biomonitoring data are based on visual scores that can be calibrated against spectral areal images (Roberge et al., 2016). Visual scores are also frequently used to assess resistance properties in common garden experiments (Newcombe and Chastagner, 1993; Albrectsen et al., 2010a; Robinson et al., 2012). Additionally, molecular detection and quantification techniques are increasingly used to analyze individual trees or small groups of trees (Feau et al., 2009; Ioos et al., 2010; Tan et al., 2010). Fungi share a conserved ITS rDNA sequence that can be amplified using specific primers, enabling PCR-based molecular verification and quantification of fungal infestations (Bourassa et al., 2005; Husson et al., 2013; Boutigny et al., 2013a; Boutigny et al., 2013b; Bergeron et al., 2019; Siddique et al., 2022). The Melampsora genus includes several species that use P. tremula as summer host including M. medusae, M. larici-populina, which are exotic in Sweden, and M. pinitorqua, which is native to Sweden. M. medusae and M. larici-populina have not been reported from northern Sweden, but since both pathogens are compatible with Populus tremula, we hypothesized that they could occur in northern Sweden for different reasons: M. medusae introduced by forestry from Canada along with fast-growing Pinus contorta and M. larici-populina through spread from distant populations of larch, potentially due to climate change. Primer pairs to detect these species are available (e.g., Bergeron et al., 2019) or can be designed using sequence data from sources such as NCBI GenBank (EU808032.1).

The aim of this study was to establish the relationship between visual rust scores and the presence of rust in leaves of aspen trees. First, we asked if the mycobiome marker genes in aspen leaves agreed among specific DNA coding for the Melampsora genus and for those of three rust species: the native M. pinitorqua and two exotic rusts (M. medusae, and M. larici-populina). Second, we asked if the impact of host genotype affected the relationship between visual rust scores and fungal DNA in the leaves. Third, we asked if expressed or genotype specific condensed tannin representations could explain the relationship between visual rust scores and the Melampsora representations.

PCR-based identification of Melampsora taxa confirmed the presence of the native pine twisting rust (M. pinitorqua) in leaves of Populus tremula but not that of the exotic rusts M. medusae and M. larici-populina. Moreover, the abundance of M. pinitorqua (quantified using a Qubit fluorometer) correlated strongly with that of the Melampsora genus in the samples. A strong positive relationship was also suggested between Melampsora-specific primer products and the products of ITS2 region of fungal rDNA amplified with ITS3f+ITS4r primers, indicating that Melampsora rust may have dominated the fungal community during the sampling event. However, there was also strong evidence that the conditions for rust establishment were highly dependent on the aspen genotype.

Fungal communities in tree foliage are not uniformly distributed or stationary (Toju et al., 2019); their composition is influenced by several factors including host genotype (Albrectsen et al., 2010b), growth site (Siddique et al., 2021), seasonal factors (Albrectsen et al., 2010b), ontogeny (Bourassa et al., 2005; Boutigny et al., 2013a), tissue type (Siddique et al., 2021), and host-associated consumers (Arnold et al., 2003; Albrectsen et al., 2010b; Siddique et al., 2021). The mechanisms underpinning these relationships are not fully understood but it has been suggested that both the host’s chemical defense responses (Ullah et al., 2017; Witzell et al., 2022) and competition between members of the mycobiome may help shape the microbial community (Wemheuer et al., 2019; Siddique et al., 2021). In the present study, we also obtained results suggesting the occurrence of a transition in the composition of the fungal community. Interestingly the genotype with the highest load of Melampsora DNA had both intermediate visual rust scores (VRS) and an intermediate level of metagenomic fungal representation based on ITS2 amplification. However, these two traits also had opposing relationships with genotype. It may be that the VRS were influenced by ontogenetic factors because low scores (VRS level 1-3, Figure 1 ) mainly reflect the presence of orange urediniospores. As the infestation progresses, teliospores start developing. In addition, the uredinosporangia generate an uneven leaf surface with disrupted underlying leaf cells, which could facilitate colonization by decomposing epiphytes. This may be why the observed relationship between genotype and ITS2 DNA abundance was opposed to that for genotype and VRS. The high amplification of Melampsora-related DNA in certain genotypes (particularly genotype 60) could thus be due to a combination of these ontogenetic processes. Overall, our results suggest that while VRS may be useful indicators of rust incidence in a stand, it does not appear to accurately predict Melampsora DNA loads. Pathogen virulence depends on the three factors comprising the disease triangle, namely the pathogen, host, and environment. However, in the case of rust, the existence of an alternate host creates complex relationships among these factors, and visible rust symptoms on aspen trees do not seem to be reliable indicators of twisting rust abundance (and thus of risks to adjacent pine trees).

Identification, detection, and quantification of disease-causing organisms are all important for biomonitoring of native and forest pathogens. Approaches used to detect and monitor pathogens include scoring of morphological symptoms, culture-based studies, and microscopic identification, all of which have their limitations (Husson et al., 2013; Boutigny et al., 2013b). PCR-based molecular techniques can detect and quantify disease-causing agents directly by amplifying their DNA (Boonham et al., 2008; Sundelin et al., 2009; Pothier et al., 2011; Husson et al., 2013; Boutigny et al., 2013a; Boutigny et al., 2013b; Yu et al., 2019; Avenot et al., 2022), and have been particularly successful in targeting and detecting known pathogens (Bourassa et al., 2005; Husson et al., 2013; Boutigny et al., 2013b; Guinet et al., 2016; Bergeron et al., 2019). Accordingly, we successfully used PCR to confirm the presence of a mycobiome and Melampsora taxa in the studied leaf samples. However, the exotic rusts M. medusae and M. larici-populina were not detected in our PCR experiments, although it should be noted that such negative results cannot be considered confirmatory (Sober, 2009). PCR-based approaches also have their downsides; notably, they only amplify a small region of a marker gene. This may limit their resolution because the targeted sequence may be common to multiple taxa, potentially leading to false identifications resulting from mismatches between the chosen primer pairs and the organism being targeted. Such problems can be overcome in part by using high throughput sequencing methods (Lorrain et al., 2018) and the risk of detection errors can be further reduced by RNA sequencing (Yu et al., 2019; Avenot et al., 2022).

Aspen is rich in phenolic defense compounds including condensed tannins (CTs), which have been negatively associated with endophytic communities (Bailey et al., 2005), the aspen pathogen Venturia (Bandau et al., 2021), and rusts (Walkinshaw, 1989). We therefore expected negative associations between foliar CT levels and fungal markers. Unexpectedly, we instead observed a positive relationship between VRS and CTs as well as weak negative correlations between CTs and the studied molecular fungal markers. While these patterns might initially appear to be inconsistent, they might be partly explained by the dynamic nature of CTs and the responses of rust to an impermeable CT barrier. CTs are concentrated in the epidermal layer of leaves (Rodríguez et al., 2022, supporting information S3), where they provide a barrier to pathogen entry. However, they do not restrict the spread of spores on the leaf surface, where multiple infestation attempts would trigger hypersensitivity responses (HR) resembling the later teliospore stages of rust infestation. CT-rich leaves could thus exhibit apparent visual symptoms of severe infestation on the surface even in the absence of such infestation. Consequently, plants exhibiting these symptoms would not have a high abundance of Melampsora DNA and would present no great risk of infestation to adjacent pine trees. Moreover, in late summer (when sampling was performed in this study), aspen leaves will have been exposed to both abiotic and biotic stress factors. Several biotic factors cause the degradation of CTs into catechin and reactive oxygen species (Ullah et al., 2017; Gourlay and Constabel, 2019), which may explain why the sampled trees did not have unusually high or low CT contents even though the genotypes in the common garden were selected to represent extremes in the natural range of CT production within aspen (Bandau et al., 2015; Bandau et al., 2021). The dynamic behavior of CTs has made it difficult to understand their function. While early studies assumed that CTs play a general protective role (Zhang et al., 2009; Yang et al., 2018; Tong et al., 2021), a growing body of evidence suggests that it is actually this dynamic behavior that is responsible for their defensive effects, and that a tree’s dynamic range or amplitude of CT-reactivity may be the real driver of its tannin-based tolerance of and resistance to pathogens.

Large-scale herbicide treatment is no longer practiced in modern Swedish forestry; instead, current efforts to avoid pine twisting damage focus on management interventions to support healthy pine trees. Pinus contorta is highly resistant to M. pinitorqua and we previously found no evidence that the M. medusae complex had accompanied lodgepole pine from Canada when it was introduced to Sweden around 60-70 years ago (Witzell, 2017). This finding is tentatively supported by the study reported herein, in which only autochthonous rust relationships were identified.

Overall, our results support the rust control strategy adopted by the Swedish Forest Agency, which focuses on resistance in pine regeneration rather than attempting to act on rust symptoms in aspen trees. While rust scoring may provide information on the incidence of rust within a forest stand, our results do not support the use of visual rust scores to assess epidemiological parameters such as infestation risk. Our findings also suggest that Melampsora rust on aspen in Sweden is dominated by the native and autochthonous pine twisting rust, and that exotic rust species associated with Lodgepole pine and larch are highly unlikely to have invaded northern Sweden.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

AS: Formal analysis (lead); project administration (lead); validation (lead); visualization (lead); writing—original draft (supporting); writing—review and editing (supporting). LM, MD: Laboratory work (supporting); writing—original draft (supporting); writing—review and editing (supporting). BA: Conceptualization (lead); data curation (supporting); formal analysis (supporting); funding acquisition (lead); investigation (lead); methodology (lead); project administration (lead); resources (lead); software (supporting); supervision (lead); validation (supporting); visualization (supporting); writing—original draft (lead); writing—review and editing (lead). All authors contributed to the article and approved the submitted version.