Introduction

Understanding the relationship between plant-viral diseases and environmental heterogeneity is important to prevent the emergence of new viruses as well as the increase of existing virusesincidence and severity. Environmental heterogeneity, defined by biotic and abiotic factors, affects infection patterns, resulting in changes in demographic processes in the virus population (Archie et al. 2009; Biek and Real 2010; Meentemeyer et al. 2012; Fraile et al. 2017).

The family Cucurbitaceae includes many important food species, such as watermelon, melon, pumpkin, squash, butternut and zucchini (Bisognin 2002). Viral diseases, especially those spread by aphids, are considered the main limiting factors for cucurbits. In Argentina, viral diseases produce yield losses and unprofitable harvests in some years (Perotto et al. 2016). This situation was also reported in other parts of the world (Abou-Jawdah et al. 2000; Coutts et al. 2011). In Argentina, cucurbits are grown from seeds of national origin in 75% of the plantings. So far there are no virus-tolerant/resistant cultivars (Della Gaspera et al. 2013), and seed exchange among producers is very common.

Potyviruses, including Watermelon mosaic virus (WMV), Papaya ringspot virus (PRSV) and Zucchini yellow mosaic virus (ZYMV), and the Cucumovirus, Cucumber mosaic virus (CMV), are the most common and prevalent virus affecting cucurbits in Argentina (Feldman and Gracia 1992; Gracia and Feldman 1986; Nome et al. 1974; Gracia 2000; Perotto et al. 2016). Moreover, these viruses are present in most cucurbit-growing areas worldwide; all of them are transmitted by aphids in a non-persistent manner and cause severe epidemics. Numerous epidemiological studies have been conducted to determine the distribution of these diseases (Alonso-Prados et al. 2003; Ali et al. 2012; Ibaba et al. 2015; Lecoq and Katis 2014; Moreira et al. 2014).

In this work we studied the cucurbit-virus association with environmental heterogeneity, using bio-meteorological variables and virus distribution. We identified the bio-meteorological variables that had the greatest impact on disease incidence. The results of the statistical analyses highlight the importance of considering the numerous bio-meteorological variables involved in the development of viral diseases.

Materials and methods

Study areas

The study was conducted at sites in the main commercial cucurbit areas in Argentina (Fig. 1). We used a Köppen climate updated classification to describe the ecosystems of the study area (Kottek et al. 2006; Reisen 2010). We considered 4 climate types: hot arid climate (BSh), cold arid climate (BWk), temperate dry winter hot summer (Cwa) and temperate without a dry season hot summer (Cfa) (Kottek et al. 2006).

Fig. 1
figure 1

Geographical location of the sampled environments (black dots) in the main commercial cucurbit areas. Location of the studied climate types: hot arid climate (BSh), cold arid climate (BWk), temperate dry winter hot summer (Cwa), and temperate without dry season hot summer (Cfa)

Full size image

Samples were collected during spring-summer (September–March) 2010–2016. An average crop cycle (from planting to harvest) was 120 days. We combined climate types with crop cycles and obtained 14 environments. The bio-meteorological variables for each environment were recorded throughout the crop cycle. The following variables were included: Minimum Temperature (Tmmin), Average Temperature (Tmmd), Maximum Temperature (Tmmax), Relative Humidity (RH), and Accumulated Precipitation (PPacum). We got the information from INTA Water and Climate Institute (Table 1).

Table 1 Average values of meteorological variables measured during of the crop cycle (120 days) in 14 environments from 2010 to 2016
Full size table

Data collection

Samples were collected from all of the 14 environments. The size of each sampled plot for each environment was different. According to the size of the plot, following a diagonal line, 10 to 50 samples were taken; in total 491 samples of squash and melon leaves were collected. We identified the samples (with and without symptoms) and processed them in laboratory.

Identification of viruses and estimation of relative incidence and relative dispersion index

The presence of each virus was determined by a double antibody sandwich (DAS)-ELISA using commercial antisera specific for the detection of WMV, ZYMV, PRSV and CMV (BIOREBA SRL Latin America, Mar del Plata, Argentina), following the manufacturers’ instructions. In addition, molecular analyses were performed to confirm serological results. To this end, specific primers were designed (Table 2) that allowed the amplification of genomic fragments of the expected sizes by reverse transcription - polymerase chain reaction (RT-PCR). 323 samples were positive (at least to one virus); The relative incidence (RI) was estimated for each virus and each environment using those 323 positive samples.

$$ RI\ \left(\%\right)=\frac{n}{N}\times 100, $$

where n is the number of positives samples for each virus and N is the total number of positive samples.

Table 2 Primers used for the detection of WMV, PRSV, ZYMV and CMV
Full size table

The Relative Dispersion Index (RDI) was estimated to study the dispersion of viruses in each environment, based on the concept of Moreira et al. (2014). The RDI estimates dispersion by comparing the variance of the observed incidence relative to the expected variance. RID values equal to 1.0 indicate random distribution, RID values >1.0 indicate aggregation, and those lower than 1.0 indicate regularity.

Statistical analysis

Samples of melon and squash were analyzed for the presence of studied viruses using a Multiple Correspondence Analysis (MCA) of categorical data.

A frequency analysis of each virus was performed for cucurbit crops in each environment.

A Principal Component Analysis (PCA) was performed to study the variability of environments and to correlate the bio-meteorological variables with relative incidence of each virus. Furthermore, the biplot allows the visualisation of the behaviour of the environments, the virus relative incidence and the bio-meteorological variables at the same time (Gabriel 1971).

A Canonical Correlations Analysis (CCA) was performed to analyze the relationship between the virus and the bio-meteorological variables by calculating unequal combinations and determine the correlation between the two sets of variables.

Finally, an algorithm with low prediction error Regression Trees (CART algorithm) was used to identify the bio-meteorological variables that had the greatest impact on disease presence. The thresholds of bio-meteorological variables that favored the increase of the presence of each virus were also established. Each virus was studied separately.

All the statistical analyses were performed with InfoStat software (Di Rienzo et al. 2018).

Results

All four viruses WMV, ZYMV, PRSV and CMV in the collected samples were found. For subsequent analyses, the data sets of melon and squash were combine. This was possible because the MCA showed that the presence of these viruses was not dependent on the host (data not shown).

Molecular and serological analysis showed that WMV had the highest RI (46%), CMV had the lowest RI (9%), and PRSV and ZYMV had intermediate RI values (20% and 24%, respectively) (Fig. 2).

Fig. 2
figure 2

Average relative incidence of the viruses Watermelon mosaic virus (WMV), Zucchini yellow mosaic virus (ZYMV), Papaya ringspot virus (PRSV) and Cucumber mosaic virus (CMV)

Full size image

All the crops were infected with more than one virus in all the studied environments. WMV was present in all the analysed environments with an incidence higher than 50%, except for BSh’11/12. ZYMV and PRSV were detected in 11 of 14 environments, with incidence ranging from 12.5 to 100%. CMV was in 7 of 14 environments, all of them being arid climates (BSh and BWk). The four viruses together were in six environments, all of them with arid climates (BSh and BWk) (Fig. 3).

Fig. 3
figure 3

Percentage of cucurbit plants infected with Watermelon mosaic virus (WMV) (black), Zucchini yellow mosaic virus (ZYMV) (gray), Papaya ringspot virus (PRSV) (light gray) and Cucumber mosaic virus (CMV) (white) viruses, in the different environments (14)

Full size image

RDI was estimated considering only infected plants. The distribution of the infected plants was random because the RDI was equal to one.. This meant that the observed incidence was the same as expected.

The PCA biplot (Fig. 4) shows that 68.7% of the total variability was explained by the first two principal components (PCs). The first PC differentiates BSh´11/12, BWk11/12, BWk12/13, BWk14/15 and BWk15/16 environments from Cfa12/13, Cwa12/13, Cfa13/14 and Cwa14/15 environments. A high positive correlation was observed between CMV and temperature variables (Tmmd, Tmmin and Tmmax). ZYMV and PRSV showed a high correlation with PPacum and RH. There was also a negative correlation between potyviruses (WMV, ZYMV and PRSV) and temperature variables.

Fig. 4
figure 4

Principal Component Analysis considering bio-meteorological variables (of the 120 days of the crop cycle), the viruses present (Watermelon mosaic virus (WMV), Zucchini yellow mosaic virus (ZYMV), Papaya ringspot virus (PRSV), and Cucumber mosaic virus (CMV), and the different environments

Full size image

The CCA scatter plot showed a high correlation between viruses and bio-meteorological variables (Fig. 5), with high support (Pearson’s r = 0.96, P < 0.001).

Fig. 5
figure 5

Scatter plot for canonical variables that represent the virus presence against canonical variable of bio-meteorological variables. Polynomial of order 1

Full size image

Temperature variables were negatively correlated with the three potyviruses, with statistical significance only for WMV (p = 0.0083).

CMV was positively correlated with temperatures. Relative humidity was positively correlated with potyviruses but was not correlated with CMV (data no shown).

The CART algorithms showed that Tmmd was the variable with most weight to explain WMV presence. At Tmmd lower than or equal to 25.6 °C, there was 0.93 probability of WMV presence in cucurbits crops; at Tmmd above 25.6 °C, the probability decreased to 0.74 (Fig. 6 up-left).

Fig. 6
figure 6

Regression trees for each virus and climatic conditions, showing percentages of incidence under predisposing climatic conditions. Watermelon mosaic virus (WMV), up-left, Cucumber mosaic virus (CMV) up- right, Papaya ringspot virus (PRSV) down- left and Zucchini yellow mosaic virus (ZYMV) down- right

Full size image

Tmmin was the most important variable to explain CMV presence. At Tmmin lower or equal to 18.9 °C, a very low presence was expected (0.05), whereas at Tmmin above 18.9 °C, probability of virus occurrence was 0.76 (Fig. 6 up- right).

For PRSV and ZYMV, RH was the variable that best predicted virus presence. For RH values lower than or equal to 58.3%, a 26% presence of ZYMV was expected, whereas for RH values higher than 58.3%, a presence of 82% was expected (Fig. 6 down- right).

For PRSV, RH lower than or equal to 61.5% and Tmmax lower than or equal to 33.5 °C, the probability of PRSV presence was 6%. In contrast, for higher RH values (61.5%), probability of virus presence was 99% (Fig. 6 down- left).

Discussion

Environmental heterogeneity can shape the plant-virus relationship, furthering the appearance of new diseases in crops, or altering their incidence and severity. The environment can modulate the plant-virus interactions, without inducing genetic changes in the virus or the host (Fraile et al. 2017).

We conducted this study in different environments during 6 years; we identified the bio-meteorological variables involved in the relationship between environmental heterogeneity and virus incidence. All viruses studied in this work are non-persistent and transmitted by aphids. These viruses can be transmitted by several aphid species, but with different efficiency (Lecoq and Desbiez 2012; Francki et al. 1979; Gal-On 2007; Alonso-Prados et al. 2003). In this work, we inferred the relationships between the viruses and environments with highly reliable results. Other works have mentioned the same relationships (Fraile et al. 2017; Kone et al. 2017; Roossinck and García Arenal 2015).

The combination of melon and squash data resulted in a higher total number of samples and provided more powerful results of the statistical analysis. WMV, PRSV, ZYMV and CMV are the most important viruses reported in cucurbits crops worldwide (Alonso-Prados et al. 2003; Ali et al. 2012; Bananej and Vahdat 2008; Lecoq and Desbiez 2012; Vučurović et al. 2012). These viruses cause serious damage that may lead to 100% yield loss and 95% decrease in the value of marketable production (Massumi et al. 2007; Lecoq and Desbiez 2012; Desbiez et al. 2009; Köklü and Yilmaz 2006; Luis-Arteaga et al. 1998; Lecoq et al. 1998; Vučurović et al. 2012). Argentina is not an exception, as WMV, PRSV, ZYMV and CMV were detected in all climates and environments over the 6 years of cucurbit monitoring.

WMV was the most common virus present in cucurbit crops in Argentina, both in arid and temperate climates. A high presence of this virus is expected even in unfavourable environments. CMV was the virus with lowest incidence, in agreement with results reported for South Africa (Ibaba et al. 2015). Some virus species (CMV and WMV) are more influenced by temperature variables, whereas others are influenced by variables of precipitation and relative humidity (PRSV and ZYMV).

Furthermore, our results show that all the environments had more than one virus, independant of the environment heterogeneity. There are about 130 genera of curcubits: however only four are of commercial importance: squash, watermelon, melon and cucumber. In Argentina, the main cucurbit crop is squash. Cucurbita moschata type (70%) and Cucurbita maxima (30%) are grown over 27,000 ha. .

The low virus genetic diversity in the all monitored environments and the presence of more than one virus suggest a low or no genotype-environment interaction. We clearly observed a high prevalence of viral diseases in all environments. The diversity and species composition of plant communities can have significant effects on virus prevalence in some productive systems (Power et al. 2011).

Viral diseases were detected in all environments but in different proportions. This difference (between arid and temperate climates) in virus proportion could be due to the different climatic conditions among environments, which also affect virus transmission, replication, incidence, and severity, and the biology of the host and/or vectors (Hull 2009; Fajinmi 2011).

The spatial pattern of viral disease provides an insight into the pathogen dynamics. A random distribution can be attributed to low biodiversity and lack of improved virus resistant/tolerant cultivars.

High temperatures during the 120 days of the crop cycle favoured CMV infection; similarly, Alonso-Prados et al. (2003) reported that CMV incidence was positively correlated with mean maximum and average temperatures. In addition, recent studies have found that high temperature results in symptoms and increases CMV titers (del Toro et al. 2017). WMV was negatively correlated with temperatures, unlike previous results (Alonso-Prados et al. 2003). This disagreement could be attributed to the different growing seasons when the studies were conducted. These results confirm that temperature is an environmental variable that affects the interaction of compatible hosts with different RNA viruses in different ways (del Toro et al. 2015).

The information provided by CART is very useful as a reference parameter. It can predict disease incidence based on prevailing environmental conditions during the crop cycle. However, it is also possible that under favourable conditions, this algorithm cannot predict the virus presence; in this case, the environment should be considered as a risk environment.

All the future analyses performed in other environments will contribute to a better understanding of plant-virus interactions with the environment. The field data obtained for this study and the analyses performed may help to predict the outbreak of viral diseases in cucurbit crops.