Root-knot nematodes (RKN) of the genus Meloidogyne are polyphagous obligate sedentary endoparasites that depend on the induction of feeding sites in the roots to complete their life cycle. Second-stage juveniles (J2) penetrate the root; migrate through the intercellular space to enter the vascular cylinder in order to induce the formation of a feeding site. Afterwards, the nematode has three consecutive moults to third (J3) and fourth (J4)-stage juveniles and to adult females. Mature females lay eggs into a gelatinous matrix attached to the posterior end of the female ( Escobar et al. , 2015 ).

Watermelon, Citrullus lanatus (Thunb) Matsum & Nakai, is susceptible to the most common RKN species, M. arenaria (Neal) Chitwood, M. incognita (Kofoid & White) Chitwood and M. javanica (Treub) Chitwood ( Thies & Levi, 2003 ; Anwar & McKenry, 2010 ; Thies et al. , 2010 ). However, physiological variability among RKN species, in terms of penetration rates, root galling severity, and final population densities (Pf) were also found among isolates of these RKN species in watermelon ( Winstead & Riggs, 1959 ; Edelstein et al. , 2010 ; Cohen et al. , 2014 ; López-Gómez & Verdejo-Lucas, 2014 ). Meloidogyne javanica showed less ability than M. incognita to form galls, produce egg masses (EM), and lay eggs in watermelon genotypes ( Cohen et al. , 2014 ; López-Gómez et al. , 2016 ).

The reproductive success of the nematode depends on the status of the host plant, population size, and soil temperatures, which in turn affect the length of the life cycle, which is slower in poor and resistant hosts than in good hosts ( Trudgill, 1995 ). The reproduction factor (Rf =Pf/Pi-initial population density), is used in Nematology as an indicator of the suitability of a host plant for the nematode. Thus, susceptible plants show a Pf/Pi >1 whereas resistant or non-hosts, a Pf/Pi <1 ( Seinhorst, 1967 ). Watermelon germplasm and commercial cultivars vary in host suitability levels as demonstrated by the Rf values in different genotypes. The M. incognita Rf ranged from 2.9 in ‘Sugar Baby’ to 9.6 in ‘Charleston 76’ ( Montalvo & Esnard, 1994 ). The Rf for M. arenaria was 2.8, 1.8, and 5.4 on watermelon ‘Crimson Sweet’, ‘Dixie Lee’, and ‘Charleston Gray’, respectively ( Thies & Levi, 2003 ). The Rf for M. incognita on Watermelon ‘Congo’ and ‘Charleston Gray’ was 7.5 and 7.0, respectively ( Pofu et al. , 2011 ). Davis (2007) reported Rf = 6.6 for M. incognita on watermelon ‘Cooperstown’ whereas in ‘Royal Sweet’ the Rf was 1.2 for M. incognita ( Xing & Westphal, 2012 ). However, cultivation of susceptible watermelon in RKN-infested fields decreased the Pf despite the presence of root galling ( Vawdrey & Stirling, 1996 ; Davis, 2007 ; Anwar & McKenry, 2010 ; Xing & Westphal, 2012 ; López-Gómez et al. , 2014 ). Reduced Pf could be due to root damage directly related to the Pi. As the Pi increases so does damage and the competition for available feeding sites which can impair growth of the nematode as well as of the host plant. In fact, reduction in plant-growth parameters is the most frequently reported outcome of the host-parasite interaction.

The leaf chlorophyll content (LCC) has been used as an indicator for evaluation of nematode damage ( Loveys & Bird, 1973 ; Melakeberhan et al. , 1985 ; Giné et al. , 2014 ; López-Gómez et al. , 2015 ) due to the strong correlation between LCC and the nitrogen content of the plant ( Gholizadeh et al. , 2009 ; Pôrto et al. , 2011 ). Changes in LCC may be detected before the appearance of disease symptoms ( Wagner et al. , 2006 ).

This research was conducted to elucidate the low reproduction of M. javanica in watermelon. The post-infection development of the nematode at progressively higher initial inoculum levels was determined by quantifying nematode traits at two time points during the life cycle. The physiological variability among populations of Meloidogyne spp. in watermelon was also determined.

Two populations each of M. arenaria (MaAL30, MaAL47), M. incognita (MiAL10, MiAL48) and M. javanica (MjAL01, MjAL39) were used to determine the physiological variability among Meloidogyne spp. populations in watermelon ‘Sugar Baby’. These populations came originally from tomato cultivated plants in plastic greenhouse ( Verdejo-Lucas et al. , 2012 ) and had been maintained in pot cultures in a greenhouse.

Watermelon seedlings were transplanted at the cotyledon stage to clay pots filled with 20 cm ³ of sterilized river sand and allowed to grow for two weeks before inoculation with 200 J2 of the respective populations. The treatments were replicated seven times. The nematode inoculum was obtained following a procedure similar to that described for the initial inoculum experiments. The experiment was repeated once. Plants were maintained in a growth chamber at 26ºC with a 16-h light photoperiod, watered daily as needed and fertilized with Osmocote® at the beginning of the experiment. Plants were harvested seven days post-inoculation. The roots were washed free of soil, stained with 0.05% acid fuchsine ( Bridge & Page, 1982 ) and the numbers of infection sites, nematodes inside the roots, and their developmental stages were determined.

The SAS system V8 (SAS Institute Inc., Cary, NC, USA) was used for statistical analyses. Prior to the analyses, when needed, nematode data were log transformed [log10 (x+1)] to homogenize the variances. Data from the repeated experiments were combined because there was no significant difference between the individual experiments. The relationship between Pi and the area under the vine-length curve (AUC) was estimated using linear regression. The AUC over time was determined using the trapezoidal method as implemented in Matlab R2102a (MathWorks Inc., Natick, MA, USA). The relationship between Pi and penetration rates and nematodes in the roots was tested by fitting an asymptotic regression approach to the data. The relationship between Pi and top dry weight and root fresh weight were also explored using an asymptotic regression curve fitting approach. Data on the number of females and on the physiological variability among RKN populations were analysed using analysis of variance (ANOVA). When the analyses showed statistical differences ( p <0.05), the means were separated according to Tukey HSD (Honestly Significant Difference) Test.

The penetration of M. javanica in watermelon roots followed a similar pattern to that previously described ( López-Gómez & Verdejo-Lucas, 2014 ). The high correlation between Pi and total number of nematodes in the roots suggests that the range of Pi values did not surpass the capacity of the root system to host the nematode. The increase in the M. javanica population densities (Rf >1) in watermelon confirmed its suitability as a host for the nematode. However, an inverse relationship was not found between Pi and Rf typically occurring in good host plants ( Seinhorst, 1967 ). The population dynamics of M. javanica in watermelon revealed low magnitudes for the maximum reproduction rate and equilibrium density ( López-Gómez et al. , 2014 ), which is in agreement with the poor-host status definition given by Seinhorst (1967) . Any trait preventing pathogen establishment or limiting their proliferation can be considered a source of resistance. In fact, watermelon shows low to moderate RKN population increases and decreased susceptibility in comparison with other susceptible crops. The Rf for M. incognita in watermelon ‘Sugar Baby’ was 2.9 and 9.6 in ‘Charleston 76` compared with 24.8 in tomato ‘Rutgers’ ( Montalvo & Esnard, 1994 ). Anwar & McKenry (2010) reported Rf= 2.5 for M. incognita in watermelon in contrast to Rf= 125 in tomato. Reproduction of M. incognita and M. javanica in watermelon genotypes was lower than in cucumber and melon ( Cohen et al. , 2014 ), as well as in hybrids of Cucurbita maxima × C. moschata used for grafting watermelon ( Thies et al. , 2015 ; López-Gómez et al. , 2016 ).

The low Rf of M. javanica in watermelon was formerly explained by a reduction in J2 penetration and a delay in nematode development in the roots during the first 11dpi in comparison with other cucurbits ( López-Gómez & Verdejo-Lucas, 2014 ). Meloidogyne javanica induced the formation of feeding sites as the presence of large numbers J3 in the roots indicated. This observation concurs with their successful establishment in the roots because the moult from J2 to J3 occurs only after the feeding site is established ( Escobar et al. , 2015 ). The establishment of feeding sites is required to satisfy the nematode nutritional demands for development and thus those sites need to be maintained throughout the life cycle until reproduction is completed ( Abad et al. , 2009 ). The dominance of J3 at 26 dpi, when egg-laying adult females are expected to have developed ( López-Gómez et al. , 2014 ) suggests the failure of the feeding sites to support nematode development beyond the J3 stage. The disruption of the life cycle at the J3 stage is reflected by the similarities in the number of egg-laying adult females at 26 and 60 dpi. Furthermore, these similarities exclude the hypothesis of delayed development as previous observations have implied ( López-Gómez & Verdejo-Lucas, 2014 ; this study). Nematodes penetrating roots but stopping their development could degrade and eventually die, leaving the galls empty ( Stephan & Trudgill, 1982 ; Faske, 2013 ; López-Gómez et al. , 2015 ) and their vestiges would not be recognized when dissecting the galls. Because the total numbers of nematodes inside the root rose with the Pi level, the low reproduction of M. javanica was not explained by crowding of the invading J2. Therefore, the key event in the host-parasite interaction was the development from J3 to J4, which was interrupted. Consequently, only a small number of the invading J2 reached the egg-laying female stage and produced offspring. These results agree with field observations on low Pf densities after watermelon cultivation ( Davis, 2007 ; Xing & Westphal, 2012 ; López-Gómez et al. , 2015 ). On the other hand, the display of profuse root galling ( Thies & Levi, 2003 ; Davis, 2007 ) but suppressed nematode reproduction suggests the hypersensitivity of watermelon to RKN, as previously noted ( Anwar & McKenry, 2010 ).

Meloidogyne javanica did not cause disease in watermelon in this study as the plant growth parameters increased instead of decreasing with the progressive increase in Pi. Indeed, watermelon compensated for the M. javanica parasitism by increasing the length of the vines (19 to 33%) and the dry top weight (40%) in comparison with the non-inoculated plants. Plants inoculated with the highest inoculums level (300J2/plant) grew faster than the non-inoculated plants. Plant growth is influenced by the nematode population density, and stimulation of top weight has been detected at low Pi values ( Wallace, 1971 ). The increased root weight, from 23 to 70% with respect to non-inoculated plants could be expected as it is characteristic of Meloidogyne infection caused by root-tissue hyperplasia and hypertrophy. Root weight of ‘Royal Sweet’ increased with the increase in Pi of M. incognita ( Xing & Westphal, 2012 ). However, in certain host-parasite combinations, the large size of the root system offsets nematode damage, as in the cucurbit hybrid rootstocks used for grafting watermelon ( Edelstein et al. , 2010 ; Thies et al. , 2010 ). The lack of differences in LCC suggests that the range of Pi tested (0.05 to 0.06 J2/cm ³ soil) were below those that may cause RKN damage and associated nutrient deficiencies in watermelon. Declines in LCC have been recorded in zucchini at higher Pi values but not lower than 1.81 M. javanica J2/cm ³ soil ( López-Gómez et al. , 2015 ).

The physiological variability typically associated with the genus Meloidogyne was noted both between populations and species of the nematode. Thus, the M. incognita populations differed in invasion rates but not those of M. arenaria or M. javanica . Meloidogyne arenaria had lesser ability than M. incognita and M. javanica to invade watermelon roots, as also shown in tomato, zucchini and cucumber ( Verdejo-Lucas et al. , 2012 ; López-Gómez et al. , 2015 ). The extent of nematode damage and the severity of the disease are affected by the physiological and genetic characteristics of the species or population infesting a given site, and thus, plant damage would change from site to site. A 5-fold increase in M. javanica Pi did not increase the Rf in ‘Sugar Baby’ and had no effect on dry top weight ( López-Gómez et al. , 2016 ). By contrast, a 10-fold increase in M. incognita Pi resulted in Rf <1 and reduced dry top weight in ‘Royal Sweet’ ( Xing & Westphal, 2012 ). The estimated tolerance limit (the nematode density below which there is no yield loss) for M. incognita infecting watermelon was 4 J2/100 cm ³ soil ( Xing & Westphal, 2012 ) whereas for M. javanica the value was 20 J2/100 cm ³ soil ( López-Gómez et al. , 2014 ). Watermelon genotypes showed less root galling when infected by M. javanica than M. incognita ( Cohen et al. , 2014 ). These results suggest that watermelon has higher tolerance to M. javanica than to M. incognita , and therefore would tolerate higher Pi levels before showing symptoms of damage and yield losses. Additional research would be needed to establish how the physiological variability in RKN can affect the host-parasite relationship in watermelon.

In summary, this study demonstrates that M. javanica development in watermelon roots was disrupted at the J3 stage. Further development of the nematode leading to completion of the life cycle was attained only by a low proportion on the invading nematodes. Consequently, M. javanica exhibited low to moderate reproductive success in watermelon. The range of Pi tested had a positive stimulating effect on plant growth parameters. These findings help elucidate the host-parasite interactions in a crop where currently no commercial watermelon cultivars are resistant to Meloidogyne .