The establishment and the monitoring of the sentinel plantation was achieved with the collaboration of the Applied Zoology and Animal Conservation group of the University of the Balearic Islands (UIB). First, the agreement of the local government and the UIB authorities had to be obtained. Then, the plant material was bought at the Calle-Plant Nursery in Wetteren, Belgium. It consisted of dormant material: 30 Salix alba 0/1 80/120, 30 Quercus petraea 2/0 80/100 and 30 Prunus domestica cv. Opal 2 years grafted on Myrobolan or St Julien. Although all the plants were equipped with a phytosanitary certificate, X. fastidiosa specific detection tests were performed on several twigs of each plant to make sure the initial material was free of the bacterium. For this purpose, three branch parts of each plant were collected and bark peeled. They were chopped and their DNA was extracted according to the CTAB-based DNA extraction protocol specific for X. fastidiosa plant samples (“PM 7/24 (4) Xylella fastidiosa”, EPPO 2019). The detection was then performed by PCR (Minsavage et al. 1994). After this double check, the ninety plants were wrapped in hessian bags filled with wood chips and were brought by truck from Belgium to the UIB campus in Palma (Majorca, Balearic Islands) in March 2018. The chips were humidified during the 2-day trip to avoid root dryness.

The location of the plot was chosen with the UIB collaborators mainly based on the ease of connection to an irrigation system, as well as on the observation of Philaenus spumarius and Neophilaenus campestris nymphs on the ground vegetation and the presence of host plants, such as wild olive and almond trees. For the positioning of the plants in the plot, the JMP software was used to generate nine blocks, each one composed of three plants of each species randomly distributed (Fig. 1). The scheme was divided by blocks to take into account the potential gradients such as the slope, irrigation distribution or sunlight. The trees were planted directly into the ground to promote the growth of the root system and to enable them to survive throughout the season (Fig. 2). The soil was compact and rocky and was dug thanks to machines (Fig. 3). In every hole, about 20 litres of breeding soil were poured. The trees were separated from each other by 1.50 m and the whole plantation covered a total area of 144 m². The irrigation system was established in the second year of the plantation. It consisted in three closed loops of pipe with one dripper per plant, allowing a constant pressure in all pipes and the same amount of water per plant (Fig. 1). The climatic data were followed through the season thanks to an HOBO device placed in the middle of the plantation.

Figure 1. 

Scheme of the sentinel plantation of Salix alba, Prunus domestica cv. Opal and Quercus petraea. The dotted lines delimit nine blocks in which there are three plants of each species distributed randomly (JMP). The solid blue line is for the representation of the irrigation system consisting in three closed loops of pipe with one dripper per plant.

Figure 2. 

The sentinel plantation in Palma (Majorca, Balearic Islands) in May 2018.

Figure 3. 

Overview of the planting of the sentinel trees. Pictures highlighting the difficulty of establishing the sentinel plantation in the compact and rocky soil of the area.

To monitor the circulation of the bacterium in the plot and around it, a 100-m demarcated area was organised around the plantation. In this area: i. a floristic inventory was carried out; ii. insect vectors were sampled; iii. a rosemary “spy plant” network was established (Fig. 4).

Figure 4. 

Surroundings of the sentinel plantation A Google Earth view (Google Earth Pro, satellite image of 6 May 2021) of the UIB campus with the location of the sentinel plantation (purple square) and the 100-m demarcated area around the plantation (yellow circle) B scheme of the plantation and the demarcated area. In the demarcated area, a floristic inventory was carried out, insect vectors were sampled in the determined quadrat and a rosemary “spy plant” network was established by planting evenly seedlings around the plantation.

As the planting of the sentinel plants with machines had removed the herbaceous layer in the sentinel plantation, which could prevent insects from reaching the trees, it was decided to re-sow grass in February of the second year to reconstitute this layer. The seed consisted of a universal mix of Asteraceae, Fabaceae and Poaceae.

Visual inspections were carried out for each sentinel tree. The appearance of Xylella-like symptoms was cautiously observed and wilting, shoot dieback, desiccation, defoliation or any change in leave colour were reported. The evolution of the size of the different plants was also monitored, as well as the presence of Xylella-vectors or of other pests or organisms. In parallel, molecular analyses were performed on each plant. One sample per plant was collected, consisting of ten leaves per plant and 4–5 small twigs collected from all sides of the plant, but prioritising symptomatic areas if there were any. DNA extractions were carried out with the CTAB-based protocol (EPPO 2019) on leaf mid-ribs, on petioles and on the twigs after bark peeling and cutting them into small pieces. The DNA samples were then sent to Belgium where they were processed by PCR of Minsavage et al. (1994) in the first three years. In the final testing of the fourth year, two samples per plant were collected, one sample consisting of 10 different twigs distributed throughout the plant together with 10 to 20 leaves, always prioritising symptomatic parts. After extraction, they were processed by PCR of Minsavage et al. (1994), as well as by real-time PCR of Harper et al. (2010). No fertiliser was applied and no pruning was carried out in the winter, to allow the plants to develop naturally and not to cut potentially infected sections.

The planning of the plantation monitoring during the four years is available in Table 1. The first year, it was decided to monitor the plantation and the demarcated area almost every month of the vector-season to assess the vector density fluctuation and to measure the rate of infection, if any, of the different plant species. In March, nymphs were sampled while, from May to October, insect adults were monitored. Several rosemary plants had desiccated already in May of the first year. Therefore, the dead ones were replaced in May and also in February of the following year. From the second year onwards, the sampling periods were chosen to correspond more or less to the beginning and the end of the highly infectious period of X. fastidiosa carried by the insect vectors, respectively, June and October, avoiding the aestivation periods of insects. The third year was impacted by the Covid-19 crisis and only one sampling campaign could be carried out in October 2020.

About 170 trees were inventoried: 134 carob trees, 31 wild olive trees, five almond trees and two pine trees. Their distribution can be observed in Fig. 5. The wild olive trees and the almond trees are both host plants of X. fastidiosa. Therefore, 36 host plants of the bacterium were identified in the 100 m around the plot (Fig. 5B). Amongst these host plants, 64% showed leaf scorching symptoms similar to those caused by X. fastidiosa (Fig. 5C). Concerning the ground vegetation, the identified plants were mainly: Conium maculatum (Apiaceae), Foeniculum vulgare (Apiaceae), Cichorium intybus (Asteraceae), Dittrichia viscosa (Asteraceae), Galactites tomentosa (Asteraceae), Euphorbia medicaginea (Euphorbiaceae) and many Poaceae (Oryzopsis sp. and others).

Figure 5. 

Tree species inventory, host and health status around the sentinel plantation A map of the different tree species in the area of 100 m around the plantation. The pink square is the experimental plot (the sentinel plantation). The green dots are for Olea europaea var. sylvestris (wild olive tree), the blue dots for Ceratonia siliqua (carob tree), the pink dots for the Pinus sp. (pine tree) and the orange dots for the Prunus dulcis (almond tree) B map of the host status of the trees located in the area of 100 m around the plantation. The green dots are the non-host plants of X. fastidiosa and the yellow dots are the host plants of the bacterium C map of the symptomatic trees located in the area of 100 m around the plantation, presenting typical X. fastidiosa leaf scorches. The green dots are the asymptomatic plants, the yellow dots the symptomatic plants and the red dots the symptomatic plants that are host plants of the bacterium. The maps were created with the QGIS software with maps from Google Earth, Imagery 2018, DigitalGlobe.

Regarding the rosemary spy plants, molecular tests carried out over four years have not detected any bacteria in the collected samples. The rosemary plants suffered from the heat and many of them died. In May of the first year, the 12 rosemary planted in the campus were already all desiccated. The following year, they were replaced, as well as six rosemary plants located in the demarcated area. However, they did not last one year. Soil tilling performed in the demarcated area by the local gardeners also removed several plants from the ground. Only 12 out of 44 rosemary plants survived the four years of the experiment. In the first year, symptoms similar to those caused by X. fastidiosa already started to appear in May and, at the end of the first season, two thirds of the plants presented these symptoms, starting with chlorosis at the tip of the leaves, which extended to all the leaf surface and turned necrotic (Fig. 6).

Figure 6. 

Rosemary health state A sampled leaves of rosemary presenting X. fastidiosa-typical leaf scorch symptoms (May 2018) B dry and dead rosemary on the field (July 2018).

Molecular tests carried out over four years have never detected any bacteria in the collected insects of the demarcated area.

During the first season, the amount of sampled insects of both species fluctuated depending on the month. This fluctuation can be observed in Fig. 7. In March, the foam produced by the nymphs could be easily observed and in total, 40 nymphs of P. spumarius (1.9 nymphs/m², mainly at nymphal stage 3–4) and 89 nymphs of N. campestris (4.2 nymphs/m², mainly at nymphal stage 2–3) were sampled (Fig. 7A, B). The nymphs of N. campestris were always found on Poaceae, while those of P. spumarius were sampled on Asteraceae (Carduus sp.), Euphorbiaceae and other herbaceous plants. At the beginning of May, local collaborators observed nymphs of P. spumarius on one S. alba plant in the plantation, as well as two adults of P. spumarius on P. domestica.

Figure 7. 

Philaenus spumarius (Pc) and Neophilaenus campestris (Nc) samples in 2018 in the 100 m area around the sentinel plantation A number of insects sampled through the different months. The striped pattern represents the nymphs and the plain pattern represents the adults B number of nymphs per m² sampled in March C number of adults per sweep sampled through the different months D number of adults per tree (wild olive, almond or carob tree) sampled during the different months.

At the end of May, the adult stage was already present and the sampling on the ground vegetation revealed less individuals than when nymphs were sampled the previous months. The number of adults per sweep was below one, with 0.04 P. spumarius/sweep and 0.03 N. campestris/sweep (Fig. 7C). In June, the herbaceous layer had dried and almost no insects were found in the ground vegetation. Very few insects were also sampled in the tree canopy. In September, more P. spumarius adults were sampled in the tree canopy (Fig. 7D); however, the number remained low with about 0.2 adults/tree. In October, new fresh herbs had grown and the highest number of N. campestris over the season was reached in the ground vegetation (0.13 adults/sweep), while a similar density as the one sampled in May was found for P. spumarius (0.03 adults/sweep).

The following years, the number of insects collected around the plantations varied between months and years (Fig. 8A, B) with a maximum in October 2020 of 0.06 P. spumarius/sweep and 0.13 N. campestris/sweep, sampled in the ground vegetation for both species. In total, four P. spumarius in October 2019, one P. spumarius in October 2020 and one N. campestris in October 2020 were found in the herbaceous layer of the sentinel plantation, showing that few insects were also circulating amongst the trees.

Figure 8. 

Philaenus spumarius (Pc) and Neophilaenus campestris (Nc) samples in 2019, 2020 and 2021 in the 100 m area around the sentinel plantation A the total number of collected adult insects B number of adults per sweep.

Molecular tests carried out over four years have never detected any bacteria in the collected samples of the sentinel plants.

Nevertheless, first symptoms on S. alba already started to appear in June of the first year (2018) with some slight necrosis at the leaf margins of some of the plants. In July of that year, 78% (21/27 plants) of the willows had slight symptoms, while in October, 96% (26/27 plants) presented leaf necrosis starting from the tip, sometimes followed by chlorosis (Fig. 9A, B). Regarding P. domestica, slight chlorosis followed by necrosis at leaf margins started to appear in July 2018 on five of the plants (Fig. 9E). In October of the same year, ten plants had slight symptoms and two had moderate symptoms of chlorosis and necrosis of leaf margins. Finally, concerning Q. petraea, first typical necrosis on leaf margins started to appear in September of the first year. In October, these symptoms were more widespread affecting 30% of the plants (8/27 plants) and consisted of typical necrosis of leaf margins with a chlorotic halo (Fig. 9C), while two plants completely died. The following years, the same symptoms started to appear on the new growing leaves, mainly on S. alba and Q. petraea. On P. domestica, typical leaf symptoms were less frequent; however, this species presented more defoliation. The second year, the extremity of the principal stem of five plum trees and five willows started to die; for the three species, stem sprouts started to grow on 1–2 plants per species.

Figure 9. 

Xylella-like symptoms on the plants of the sentinel plantation in October of different years A on Salix alba in 2019 B on Salix alba in 2018 C, D on Quercus petreae in 2018 E on Prunus domestica in 2018.

The summer of 2021 was declared the warmest recorded in Europe in the last 30 years, with severe heatwaves in the Mediterranean (Copernicus 2022). While the sentinel plants were already weakened by the last three hot summers despite the irrigation system, many of them died completely or partially this last year. Death was assigned after scraping the bark from several parts of the trunk. In total, 14 S. alba plants were completely dead and 13 had their main stem completely desiccated, but had developed sprouts at the bottom that were still living. The remaining leaves showed all symptoms of necrotic and chlorotic leaf margins. Three Q. petraea died and almost every remaining individual presented symptomatic leaves, while two of them had their main stem completely dead, but with living sprouts. Finally, two P. domestica died and about twenty of them had symptomatic leaves, which consisted of leaves turning red from the margins with a degraded colour, except for some leaves where the discoloured margins were quite delimited. About fifteen plants had between a quarter and a half of their main stem completely dead starting from the tip. Finally, the stem of two of them had completely died, leaving a second plant to grow from the variety Myrobolan, as the Opal variety was grafted on to this one. The size measured each year was not reported here because it was biased by the death, or partial death, of the main stem.

Concerning Q. petraea, damage caused by the herbivore Lachnaia sexpunctata Scopoli (Fig. 10A) in May-July 2018 forced us to put their foliage under a net (Fig. 10B) until mid-July to keep them alive, but this also resulted in their inaccessibility to X. fastidiosa insect vectors. A pesticide (Cypermethrin 10 ml/l) also had to be applied. The following years, the situation was better and the foliage could be exposed to the environment for all seasons. During the monitoring, fungal-like agents were also observed on the leaf surface of many individuals.

Figure 10. 

Herbivore damage on Quercus petraea A Lachnaia sexpunctata feeding on Q. petraea in the sentinel plantation B net on Q. petraea to avoid the herbivores eating their foliage.

Despite the publication of EPPO (2020) providing guidelines for sentinel studies, only two other assays that describe themselves as sentinel plantations have been reported in literature and both as part of the same project (Roques et al. 2015; Vettraino et al. 2015), while a third study can be characterised as one, even if it does not refer as such (Rathé et al. 2014). The sentinel plantations of Roques et al. (2015) and Vettraino et al. (2015) consisted of a four-year monitoring of five European tree species, including Quercus spp., which had been planted in China to investigate potential new host-pest/pathogen associations that could emerge in Europe through plant trade. While the experiments allowed the collection of valuable data and discovery of new associations, it already highlighted the complexity of the technique in terms of logistics and workload.

In our study, many constraints were faced and are reported in Table 2 with some perspectives on how the system could be improved to ease the implementation of the method. Our burdens started with permits and Italian administrations. In fact, the initial plan was to establish the plantation in the Apulian area where the first epidemic was declared. Ex-patria sentinel plantation studies require the movement and planting of non-native plants and they are, therefore, subjected to the host country’s legislative and administrative procedures for importation and planting (EPPO 2020). After more than one year of back-and-forth e-mails to get approval from the Italian authorities, our request was transferred to our first correspondent. Therefore, the location of the plantation was changed to Majorca, where a good collaboration with UIB allowed us to obtain the agreement of the local authorities and the university, where the plantation was to be established, in about a month. A comparative view of the full procedural pathway between our first attempt in Apulia and Majorca can be viewed in Suppl. material 1. In addition, for administrative reasons, Roques et al. (2015) were unable to establish their plot in the initially optimal climatic zone where they wanted. In their study, many plants were lost due to the delays in Chinese authorisations and imposed quarantine measures. Due to the common external border, plants with a European passport can circulate in Europe without restrictions and sentinel plantation intra-Europe should, therefore, be easier to implement (Vettraino et al. 2020). Furthermore, Vettraino et al. (2020) classified Europe as having low bureaucratic complexity concerning sentinel plantations compared to other non-European countries in a ranking they established according to the country’s bureaucratic procedures. Surprisingly, Italy was considered the least complex European responding country, in contrast to what was experienced here. However, the current sensitive issue of X. fastidiosa in Italy has certainly not helped to speed up the procedures. On the other hand, the government of the Balearic Islands immediately accepted our request under certain conditions, which were the compliance with the norms in force in the territory regarding X. fastidiosa and the prohibition of planting Polygala myrtifolia, initially chosen as a spy plant for its high susceptibility to the bacterium. Vettraino et al. (2020) reported that most of the countries have restrictions on the import of certain plant species or genera, for example, Roques et al. (2015) were prohibited from planting Pinus spp. for their sentinel plantations in China. Finally, it is worth noting that we were not able to import plants collected in semi-natural environments, such as cuttings of S. alba, because of the difficulty of obtaining a phytosanitary passport for this type of material and all imported plants had to be purchased from Belgian nurseries in order to be certified.

The second challenge of this plantation was to keep the plants alive. The fact they were grown in an environment with different conditions including temperature and soil, brought different biotic and abiotic stresses. The life of these plants depended once again on the good cooperation on site. For example, the delay in the irrigation system establishment in the first year led the local staff to water the plants by hand every two days, carrying more than 80 litres of water in cans to the plantation. Furthermore, if they had not placed mesh covering the foliage for the herbivore L. sexpunctata that devoured the oak leaves, the plants would have died during the first year. However, despite constant monitoring by local collaborators, plant mortality increased from year to year and stress often led, especially in willows, to death of the main stem and the growth of new shoots at the bottom of the plant. This may have an impact on the outcome of the experiment, as the death of the potentially contaminated plant parts would lead to the death of the bacteria itself.

Here, the hurdles faced in sentinel plantation assays were coupled with the difficulties often encountered in X. fastidiosa studies. In fact, this bacterium is known to be fastidious for research including in its detection (Wells et al. 1987). Its concentration in plants and insects could be below the detection threshold of the different methods (Cruaud et al. 2018; EPPO 2019) and it is irregularly distributed in plants so may be missed during sampling, especially in asymptomatic plants (EFSA PLH Panel 2015; EPPO 2019). On the other hand, symptoms are not always reliable as they can easily be confused with symptoms triggered by other factors, such as drought (EFSA PLH Panel 2015). Therefore, it is more than likely that other causes, such as drought or soil stress, were responsible for the typical chlorosis and necrosis of the leaf margins observed on all three species in this study, especially for such plants used to colder temperature and more humid soil, even with the irrigation provided. While an undetectable low bacterial concentration can be questioned, several studies reported that high symptomatic responses were correlated with high bacterial loads (Holland et al. 2014; Saponari et al. 2017) suggesting a greater probability of detection if symptoms were due to X. fastidiosa infection.

Another parameter to consider when studying host susceptibility of X. fastidiosa is that the incubation period can be measured over months and years (EFSA PLH Panel 2019), indicating that time is a key element. For example, the survival time of Majorcan almond trees from bacterial infection to tree decline has been estimated around 14 years (Olmo et al. 2021). Sentinel plantation studies are already by themselves long-term assays and superimposing the potential time required for infection of the bacteria gives us an idea of how long it takes to conduct this type of experiment. However, the longer incubation period does not necessarily mean lower susceptibility to the bacterium itself, since many external factors can influence it, for example, the vector population. In fact, as X. fastisiosa is an insect vector-borne pathogen, its circulation and infection will depend upon the abundance, host preference and prevalence of its insect vectors, which are adding complexity to the system compared to other sentinel studies that would, for example, measure the direct impact of herbivores on leaves. Moreover, a particularity of diseases caused by X. fastidiosa is the polymorphism of the pathosystems. In fact, different strains and bacterial subspecies will act differently with the various xylem-feeding insect species and the different host species or cultivars, leading to very specific epidemics around the world (Pierce’s disease, Citrus variegated chlorosis, Olive quick decline…) to almost no symptoms or to an endophytic presence. While the choice of the region in relation to the strains one wants to study is essential, this means that an absence or an endophytic interaction does not mean that other strains cannot be aggressive on the same plant species and cultivar. This means that there will only be an answer for a potential pathosystem related to the chosen region, but there are multitudes of other possibilities. The identified pathosystem will keep the adjective “potential” until the disease is not actually observed in the country of origin, as local environmental conditions or the presence of an effective vector will also have an impact.

A final element to be taken into account in the case of sentinel plantations with X. fastidiosa is the European regulation as a quarantine agent (Council Directive 2000/29: EC 2000) and the European containment and eradication measures imposed in case of detection (Commission Implementing Regulation (EU) 2020/1201: EC 2020) with the establishment of a demarcated area delimiting an infected zone of at least 50 m and a buffer zone varying in terms of kilometres depending on the situation. In the infected zone, eradication measures have to be undertaken consisting of the removal of all specified host plants of X. fastidiosa. However, in areas in which the bacterium is considered widely established including Apulia, Corsica and Balearic Islands, lighter containment measures may be implemented as eradication is no longer considered feasible. Nevertheless, these measures still imply the removal of all the infected plants in the 50 m zone and an intensive surveillance within an area of at least 5 km radius together with vector control. These measures mean that, even in containment zones such as the Balearic Islands, the detection of an infected plant in the sentinel plantation would lead to the control of vector population in the area and to a decrease in the infection pressure around other plants of the plantation. Similarly, if the tested positive plant has to be removed immediately, the observation of symptom evolution and thus, the assessment of susceptibility is compromised, unless exceptional permits for scientific research are obtained. In this study, the problem did not arise because all plants tested negative. Nevertheless, we were still impacted by the consequences of the European legislation as, under the containment scenario in the Balearics, local government and UIB authorities did not advise systematic test of the host plants on the campus. In fact, a positive detection would have led to the uprooting of the campus vegetation, including, as mentioned before, our plantation if special permits were not issued. These measures are considered highly severe for an area where the bacterium is widespread and separated from other regions by the sea (Olmo et al. 2021). In areas infected by X. fastidiosa, the possibility of not having to remove infected plants in the field for scientific research purposes deserves further exploration in terms of PRA and bureaucratic procedure. Finally, for biosafety reasons related to quarantine organisms, plant samples cannot be moved and have to be processed on site, which again requires a good logistic, local collaboration and proper infrastructure.

The implementation of a sentinel plantation when studying a specific pest or pathogen requires knowing well the epidemiology of the exact spot of the establishment, as local environmental components have a great impact on the outcome of the experiment (Kenis et al. 2018). The location chosen for this study was probably not optimal, as it was later evidenced that X. fastidiosa infection pressure was low and, thus, this certainly constitutes the main reason for the lack of positive detections in insects, spy and sentinel plants in the plot. When the plantation was established on the UIB campus, the prevalence and the epidemiology of the outbreak on the island were not yet well known, which is still the case in several regions where X. fastidiosa has recently been detected. Positive detections were reported on the campus about a hundred metres from the plantation on one R. officinalis plant and two olive trees (M. A. Miranda, personal communication) and the health state of host plants including declining almond trees, one of the main crop affected by X. fastidiosa on the Island, led us to suspect that the place was infected. However, due to the lack of systematic sampling after the declaration of the contention scenario in the Balearics, the presence of the bacterium could not be confirmed by testing. In addition, the quantity of nymphs sampled when choosing the location was 1.9 nymphs/m² for P. spumarius and 4.2 nymphs/m² for N. campestris in March, which is actually higher than the mean observed in the ground vegetation sampled through the Island. López-Mercadal et al. (2021) reported an average of about 0.22 nymphs/m² for P. spumarius in the peak of March and 0.005 nymphs/m² for N. campestris with differences between plots and years. In our plot, the prevalence of these nymphs was null. However, this information was not relevant as the infectivity is lost with each moult (Purcell and Finlay 1979) and prevalence, therefore, has to be measured on adult insects to have robust data.

After deepest outbreak investigations, it appeared that the east side of the Island towards Manacor was probably the most infected part, while the plantation was located to the west side of the Island. In fact, Gutiérrez Hernández and García (2018) mapped the positive records of X. fastidiosa detected in the Balearics by the Plant Health Section of the Department of Environment, Agriculture and Fisheries of the Government of the Balearic Islands and showed that most of the positive samples were concentrated on the east side with the highest densities in agricultural and residential areas close to the main communication routes. They stressed, however, that the conducted sampling strategy could have biased this distribution, for instance, because the samples could have been collected preferentially in these more accessible areas. Based on direct field observations and using Google street view, Moralejo et al. (2020) also mapped the distribution of Xylella-symptomatic almond orchards and their mortality across the Island, tracking their evolution since 2012 (Fig. 11). They showed a gradient from east to west, showing a moderate incidence on the site of the plantation. However, molecular testing of infected almond trees did not reveal a clear spatial pattern (Moralejo et al. 2020). In addition, highly variable incidence was encountered in different orchards (Olmo et al. 2021), hence the need of knowing the incidence and prevalence of vectors at the precise location of a sentinel plantation.

Figure 11. 

Almond leaf scorch incidence through Majorca. Map of the incidence of the almond leaf scorch disease and almond mortality within orchards across Majorca in 2012 and 2017 through field observation and Google view archives according to fig. 2 in Moralejo et al. (2020) (courtesy of E. Moralejo). The figure was adapted by adding the pink star at the localisation of the sentinel plantation.

The density and prevalence of insect vectors are one of the drivers of X. fastidiosa infection and impact the temporal dynamics of symptom appearance (EFSA PLH Panel 2019), as multiple and independent infections could lead to an injection of a higher bacterial load and a decrease in the incubation period (Daugherty and Almeida 2009). The damage in the Balearics are the consequence of almost 20 years of infection (Moralejo et al. 2020), suggesting that the infection pressure could be too low to conduct sentinel plantation experiments. In fact, the abundance of nymphs and sampled adults, as well as the prevalence of insects are lower than the values encountered in the infected areas of Apulia where the outbreak was more drastic. A prevalence of 23% was reported in Majorca (López-Mercadal et al. 2021) compared to up to 71% detected in an Apulian olive grove (Cornara et al. 2016a). Similarly, higher densities of vectors were measured in Apulia with 7 to 39 nymphs of P. spumarius/m² in olive orchards (Bodino et al. 2019), about 7 adults/olive trees and 0.5 adults/sweep in weeds recorded during the respective seasonal peaks (Cornara et al. 2016b), however, with heterogeneity identified amongst the orchards studied (Bodino et al. 2019). In our plot, the adult density varied according to the seasonal estivation and ground drying pattern of Mediterranean regions (Cornara et al. 2016b; López-Mercadal et al. 2021). It barely reached a maximum of 0.04 P. spumarius/sweep in May and 0.2 P. spumarius/tree in September 2018, while the average reported through the Island was below 0.1 adults/sweep in ground cover, tree canopy and border vegetation (López-Mercadal et al. 2021). In addition, N. campestris was not considered as a significant vector due to its very low presence on the tree canopy (López-Mercadal et al. 2021). Moreover, the soil around the plantation was ploughed almost every year, as common management on the Island, which, besides destroying several rosemary spy plants, probably decreased insect movement around the plot even with sowing of ground vegetation the second year. In fact, tillage is a technique of vector control reducing the number of vectors/m² (Bodino et al. 2019; EFSA PLH Panel 2019). In the study of Kenis et al. (2018), their plantation located at the edge of the forest took less time to be infested than another one situated in an agricultural-peri-urban area, highlighting again the impact of high local circulation of pests and pathogens on the time and outcome of the assay.

Even in locations with high infection pressure, the efficiency of the sentinel plantation in the case of X. fastidiosa host range investigation is questioned due to the ratio results/time-workload. Yet the sentinel plantation method is currently being used in Apulia for the screening of olive cultivars coming from various Mediterranean olive-growing areas (Spain, Tunisia, Greece etc.) by exposing them to the natural pressure of inoculum in heavily-infected fields (XF-ACTORS 2017; Saponari et al. 2019b). The previous finding of the mild symptoms on the Leccino and FS17 olive cultivars adjacent to severely-affected orchards motivated the study (Boscia et al. 2017). Approximately 100 different genotypes were planted and are currently under evaluation in different plots, actually making the Apulian region home to one of the largest sentinel plantations of all time. This study is promising and is considered necessary for long-term management of X. fastidiosa in olive-growing regions as preliminary data show already differences in susceptibility in various cultivars (EFSA PLH Panel 2019; Saponari et al. 2019b). However, it highlights the long-term commitment required as the survey started in 2015 and is still ongoing. The project is part of a research programme funded by the European Union’s Horizon 2020 Research and Innovation Programme, which explains how a project of this magnitude could be established and which underlines the need for long-term consistent international support for the implementation of such experiments. The success of this plantation, in addition to the selection of highly-infected plots, also comes from the fact that the tested potential hosts are related to the STs present in the environment. As the Apulian ST53 is highly aggressive on olives, it is obvious to carry out olive plant susceptibility in this area. However, other X. fastidiosa infected regions, such as Balearic Islands and Corsica, could be interesting to study the susceptibility to other STs, as three STs belonging to two subspecies co-exist in Majorca, while only one in the Apulian Region.

Thus, the Apulia study proved the usefulness of sentinel plantations in the context of X. fastidiosa. However, it would be less relevant to conduct these studies in certain situations. There should be, for example, similarities between the climatic conditions of the two regions involved in the sentinel studies to minimise the impact of external factors. So far, the bacterium has only been found established in southern Europe, in regions with a Mediterranean type of climate and these studies would, therefore, be less suitable for northern European countries, as differences in environmental conditions could lead to weakening or even death of the plants and to misidentification of the cause of potential symptoms. Nevertheless, this tool remains very valuable and should be considered for studies on X. fastidiosa, as other techniques for screening potential hosts of this pathogen are also discussed. Amongst these techniques, mechanical inoculation shows a low rate of success, even in susceptible hosts (Prado et al. 2008; EFSA PLH Panel 2019) as this method artificially reproduces infection while in the environment and only xylem-specialised insect vectors have the capacity to infect plants (Almeida et al. 2005). Working with insect vectors is, therefore, a more relevant way of conducting experiments. However, besides the biosafety risk it could represent for Xylella-free regions and the need for proper infrastructure, the very act of infecting an insect is a challenge. Other experiments consisting in grafting more than 400 olive genotypes on infected trees were conducted in parallel with the sentinel plantation in Apulia to short incubation period and time imposed by insect traits (Saponari et al. 2019b). However, in addition to also being an artificial way of infection, it requires the availability of appropriate infected graft material. Therefore, sentinel plantations have their advantages and have to be considered as valuable complementary tools in certain situations.

In these situations, this study has provided a complete methodology to monitor the bacterium circulation through the sentinel plants. The use of spy plants is certainly useful if sampling of susceptible vegetation is not possible in the nearby area. In other cases, sampling of local flora may be sufficient, although it does not ensure real-time circulation of the bacteria, as the current state of the local flora could be the result of infection from the past (Moralejo et al. 2020). The use of small perennial plants may facilitate sampling, as bacteria are distributed irregularly in the plant. The species or mix of species must be adapted to local conditions, susceptible to the bacterial strains being investigated and favoured by local vectors. In this study, R. officinalis was chosen as it was reported infected with the European STs of subsp. multiplex and subsp. pauca (ST6, ST7, ST53, ST80, ST81 and ST87) and was found infected in Majorca with the ST81 (EFSA 2022). In addition, in America, the bacterium was detected on this plant species close to X. fastidiosa subsp. fastidiosa-infected vines (Freitag 1951).

Sentinel studies can also be carried out differently to study host range in countries that cannot match closely the environmental conditions of the potential location. First, arboreta and botanical gardens are still an option for studying exotic host range in naturally-infected environments. However, as a detectable infection depends on the density and prevalence of X. fastidiosa insect vectors (Daugherty and Almeida 2009), the use of this method could also be discussed as these areas are often subjected to phytosanitary management. One advantage of these studies regarding X. fastidiosa would be that plants are grown in these sites for a long time, increasing the success concerning potential latent periods or low bacterial load potentially enabling detection. In addition, the study of Groenteman et al. (2015) has shown promising results for X. fastidiosa research by sampling in botanical gardens. They managed to discover 28 New Zealand plant species infected by X. fastidiosa, including several visited by the insect vector Homalodisca vitripennis Germar, in Californian botanical gardens where the disease is well established. They also found parasites capable of controlling the vector on these plant species with the aim of a biocontrol early-response strategy in case H. vitripennis invade New Zealand.

A second way would be to carry out transmission experiments in an insect-proof greenhouse with naturally-infected vectors in contaminated regions to bypass the problems of biosecurity imposed by Xylella-free areas and the difficulty of infecting insects. Compared to standard sentinel plantations, these experiments allow us to reduce the dependence on vector density and on insect feeding preferences. In fact, although P. spumarius is considered a polyphagous species and was observed feeding on the three studied sentinel plants in their area of distribution, it is possible that, in the sentinel country, these insects are more interested in native vegetation. Native plants could, therefore, compete with the exotic sentinel ones, potentially resulting in fewer vector feeding events decreasing the bacterial transmission probability. Even if vector preferences are biased and that natural conditions are, therefore, not fully met, these experiments can still be considered as sentinel studies since they consist in ex-patria plants sent to study the impact of exotic organisms in areas in which they occur. This has been done in Majorca as a complementary experiment where 20 new cuttings of S. alba and of P. tremula have been sent from Belgium to the UIB campus. There, transmission experiments with naturally-infected P. spumarius were conducted in an insect-proof greenhouse and revealed positive infection on S. alba, proving the higher efficiency of the technique compared to sentinel plantation.

Finally, in-patria sentinel plantings (Eschen et al. 2019) or sentinel nurseries (sensu Vettraino et al. 2017) consist of planting native traded plants without phytosanitary treatments on its own land to monitor pests and pathogens which could be spread through international trade (Vettraino et al. 2017). They obviously do not have the same objective as ex-patria plantations that inform PRA of organisms that are not yet present in a given area. Rather, they consist of surveillance for a known pathogen for which possible entry and dispersal pathways have been identified (Mansfield et al. 2019) and they still represent valuable sentinel assays to be conducted with the aim of early detection of X. fastidiosa in new regions. The major difference with a standard commercial nursery is that no pest control measures are implemented on these plants (EPPO 2020), so that it is possible for the vectors to reach the plants and for the plants to become infected if X. fastidiosa is introduced in the area. For this strategy to be effective, these plantations have to be established in strategic locations where the bacterium is the most likely to enter. The “plant for planting” pathway being the main entrance for exotic organisms including X. fastidiosa (Liebhold et al. 2012; EFSA PLH Panel 2018), their locations in/close to nurseries or other plant commercial places, would be relevant. In addition, these plantations must consist of known host plants that have a high probability to be the first infected when the bacterium enters an area and, if possible, to be highly susceptible for the infection to be visible and easily detectable. For example, the Auckland Botanic Garden has set up a sentinel plot of myrtle plants to detect the potential arrival of the myrtle rust (Puccinia psidii Winter) as early as possible in New Zealand, as the fungus was prevalent in Australia at the time (Barham et al. 2015). Similarly, one can imagine planting a network of P. myrtifolia near nurseries, previously tested for innocuity, which are regularly monitored for potential contamination by X. fastidiosa. Obviously, these susceptible plants should be tested carefully and regularly to provide the benefits of early detection while preventing them from serving as inoculum for disease establishment (Mansfield et al. 2019).