Introduction

Watermelon [Citrullus lanatus (Thunb.) Matsum.& Nakai]is a Cucurbitaceae, originated in the tropical regions of Africa.It is cultivated throughout most of Brazil due to favorablesoil and climatic conditions (^{Tosta et al., 2010}). The currentwatermelon average yield in Brazil is 23.5 Mg ha^-1. Themidwestern and southeastern regions are the main productionareas in the country, and the states of Ceará and Goiásare the major producers (^{IBGE, 2013}).

According to a recent agricultural census, 603,015Mg of watermelon were produced in north-eastern Brazilin an area of 28,436 ha, representing an average yield of21.3 Mg ha^-1. Paraiba State produced 5,100 Mg in anarea of 266 ha, with an average yield of 19.2 Mg ha^-1(^{IBGE, 2013}). However, even under favorable climaticconditions for the cultivation of watermelon, primarily insemiarid areas, Paraiba State is eighth in volume producedand seventh in earnings among the north-easternstates. Therefore, watermelon cultivation should be encouragedthrough subsidies and government investmentin research that aims to improve cultural practices and toprovide information to producers for strengthening cultivationto increase earnings.

Among the factors contributing to the low productivity ofthe vegetables, including watermelon, there is lack of sufficientinformation on crop management by farmers, especiallyin respect to mineral and organic fertilizers (^{Leão et al.,2008}). Fertilization is a crucial factor for short-cycle vegetablecrops with a high demand for nutrients like potassiumand nitrogen (^{Andrade Junior et al., 2006}). Some studiesshow that watermelon is highly demanding of potassium, thenutrient that is the most absorbed and exported to the fruitsduring the productive phase (^{Grangeiro & Cecílio Filho, 2006};^{Aguyoh et al., 2011}).

There are concerns about the use of chemical fertilizers,but supplying organic matter to the soils, especially manure,promotes growth of vegetables due to its positive effects onsoil chemical, physical and biological properties. Theseeffects include increasing the supply and availability of nutrients(^{Higashikawa, Silva & Bettiol, 2010}; ^{Müller et al., 2014};Silva & Menezes, 2007), improving water storage and soildrainage (^{Hussein, 2009}), and improving the efficiency andutilization of mineral fertilizers (^{Guppy et al., 2005}). However,elevated levels of manure and other organic materials inthe soil can lead to an imbalance of nutrients, and to diseasesresulting from excessive soil moisture (^{Oliveira et al.,2007}).

The provision of adequate levels of manure to the soil,alone or in combination with chemical fertilizers, had resultedin increased yields of watermelon and other vegetablecrops like pumpkin, cucumber, and gherkin (^{Abul-Soud,El-Ansary & Hussein, 2010}; ^{Leão et al., 2008}; ^{Oliveira et al.,2009}; ^{Onyia et al., 2012}; ^{Sarhan, Mohammed & Teli, 2011}).However, because of the environmental specific conditionsin each region, the sources and amounts of organic matterapplied to the soil aiming at the increase of the productionshould be studied to prevent damage to the environment andcrop failure. The aim of this experiment was to evaluate theinfluence of the addition of various levels of livestock manureto the soil, with and without potassium fertilization on watermelonproduction.

Material and Methods

The experiment was conducted on Campo Comprido(7°04’27.1"S and 37°19’05.2"W), Patos City (Altitude = 242m), Paraiba State, Brazil, from October to December 2012.According to Koeppen classification, the region’s climate istype BSh, semiarid, with annual average temperatures exceeding25 °C, average annual rainfall < 1000 mm y^-1, andirregular rainfall (^{Ministério da Agricultura, 1972}). The localsoil was classified as NEOSSOLO Fluvico (^{Santos et al.,2013}). Soil samples were collected at 20 cm depth to determinephysical and chemical soil characteristics accordingto the procedures proposed by ^{EMBRAPA (2011)} (Table 1).

Table 1 Chemical and physical soil attributes of the soil at 20 cm depth. 

Treatments were arranged in a randomized block designwith four replications in a factorial scheme 2 × 5 + 1,referring to absence and presence (15 kg ha^-1 K₂O) of potassiumfertilization and five doses of livestock manure (0,360; 1080; 1800 and 2510 g hole^-1) in soil with nitrogen fertilization,and an absolute control treatment (no organic normineral fertilizer). The plots were composed of three lineswith seven plants. All plants of the plots were considered forevaluation of production.

For treatments with potassium, the dose (15 kg ha^-1 K₂O)was defined according to the recommendation of AgronomicInstitute of Pernambuco (IPA, 2008) for watermeloncrops. A dose of 30 kg ha^-1 K₂O is recommended in soilswith potassium content > 0.30 cmol_c dm^-3 (IPA, 2008), but asthe experimental soil area had 0.40 cmol_c dm^-3, half the recommendedlevel was provided. Potassium was suppliedin a single application, 40 days after germination, using potassiumchlorate (KCl). In addition to the ten treatments withthe 2 x 5 combinations between potassium and cattle manure,an absolute control treatment with no mineral no organicfertilizer was included. The treatments with manure andpotassium also received nitrogen in the form of urea (45 %N) at 26.68 and 80 g plant^-1 (66,7 and 200,1 kg ha^-1) at 20and 40 days after germination, respectively, as recommendedby IPA (2008). Phosphorus was not supplied becausethe soil had very high phosphorus content (Table 1).

The manure used in the experiment was chemicallycharacterized (Table 2). Manure was applied 30 days beforesowing to increase the content of soil organic matter from0.75 % (7.5 g kg^-1), the content prior to the experiment (Table1), to 1.5, 3.0, 4.5, and 6.0 % (7.5; 15; 30; 45 and 60 g kg^-1).Considering that only part of the cattle manure is carbon(380.54 g kg^-1) the doses (g hole^-1) were obtained using thefollowing expression:

D_CMA= [(D_OMA - D_OME) x Vc x Ds] /OMCM

where:

D_CMA - dose of cattle manure to be applied per hole (g hole^-1);

D_OMA - dose of organic matter to be achieved in the soil, (g kg^-1);

D_OME - dose of organic matter existing in the soil, 7.5 g kg^-1(Table 1);

Vc - mean volumetric capacity of the hole, 12 dm³ (0.20 m ×0.20 m × 0.30 m);

ds - soil density, 1520 g dm^-3;

OMCM - organic matter content in the cattle manure, 380.54g kg^-1 (Table 2).

Table 2 Chemical attributes of the manure used in the experiment. 

The watermelon cultivar used for the experiment was‘Crimson Sweet’. Planting holes (0.20 m × 0.20 m × 0.30 m)were dug 30 days before seeding, spaced 2.0 m × 2.0 m.Three seeds were sown in each hole, and when plants hadthree pairs of leaves, permanent thinning was performedleaving only one plant per hole. The experiment was irrigateddaily from sowing to germination (eight days), and everytwo days from germination to fruiting (70-75 days) by sprinklerirrigation, providing the equivalent of 24 L plant^-1 day^-1 asrecommended by ^{Carvalho (2005}). After harvest onset, fruitswere collected once or twice a week, and counted and weighedto determine the cumulative number of fruits per plant,average fruit weight (kg) and productivity (Mg ha^-1).

Data were subjected to analysis of variance (ANOVA).Means values referring to potassium fertilization were comparedby Tukey’s test (5 % probability level) and meansreferring to cattle manure were submitted to regression tests.Contrast analysis and Dunnett’s test (p < 0.05) between thetreatments with fertilization (livestock manure and mineralinputs) and control treatment (without mineral fertilizer ormanure) were performed.

Results and Discussion

The average per-plant fruit production was not affected bymanure and potassium supply to the soil according to ANOVAanalysis (p < 0.05). There was a significant effect for thecontrast between treatments with potassium fertilization andlivestock manure against the control treatment (Figure 1).The average number of fruits per plant was 0.66 in the treatmentswithout mineral fertilizers and manure (control treatment)and 1.42 fruits in the treatments with combinationsmanure x potassium fertilizer, that is, a superiority of 115 %in relation to the control treatment.

Figure 1 Number of watermelon fruits per plant in soil fertilizedwith potassium and livestock manure (factorial,■) and withoutfertilization organic or chemical (control, 􀂅). Different letterson top of each column means significant differences(p < 0.05) according to Tukey’s test. 

The supply of potassium and nitrogen to the soil in additionto manure stimulated fruit production. These two nutrientsare required in large amounts by watermelon (^{Grangeiro& Cecílio Filho, 2006}), and therefore plants respondedpositively to their supply. ^{Leão et al. (2008}) evaluated theeffects of applying manure and NPK fertilizer on watermelonproduction, and obtained 1.15 fruits per plant by providing9 L manure per hole, treatments with fertilization. An increasednumber of fruits per plant due to increased soil organicmatter was recorded in other cucurbits, such as gherkin(Cucumis anguria L.) (^{Oliveira et al., 2009}) and cucumber(Cucumis sativus L.) (^{Hussein, 2009}; ^{Onyia et al., 2012}).In pumpkin plants, ^{Sarhan, Mohammed & Teli (2011}) foundthat sheep manure significantly increased the number offruits. Comparing the treatments with chemical and organicfertilizers with the control treatment, treatments 1, 4, 8 and 10showed no significant difference (Table 3).

Table 3 Comparison between fertilized treatments and control by the Dunnet’s test (significant at the 0.05 level) for fruit valuesper plant of watermelon (kg fruits plant^-1). 

The average weight of watermelon fruits was significantlyinfluenced by manure and potassium supply (Figure 2A).For treatments without added potassium, regression modelsdid not fit for organic amendment, with an average7.0 kg fruit^-1 across manure dosages. On the other hand, intreatments with potassium applied, a lineal relationship wasestablished (w = 6.3783 + 0.0277**x), where values increasedfrom 6.37 to 8.6 kg fruit^-1 in soil received the highestdose of manure (2520 g hole^-1), which means an increaseof 35 % in absolute values.

Figure 2 (A) Average fruit weight of watermelon cultivated under livestock manure levels in soil without (-) and with (- - -)potassium supply and (B) comparison between treatments with organic or mineral fertilization (■) and control treatment (􀂅).Different letters on top of each column mean significant differences (p < 0.05) according to Tukey’s test. 

^{Leão et al. (2008}) also found that the average mass ofwatermelon fruits increased as increased manure in thesoil, reaching 5.3 kg fruit^-1 with 9 L manure per hole. Inpumpkin experiments, ^{Abul-Soud, El-Ansary & Houssein(2010}) and ^{Santos et al. (2012}) also observed an increasein average fruit weight when levels of livestock manure andliquid pig manure were applied to the soil. However, melonplants growing in soils containing different sources and levelsof organic matter including manure, ^{Souza et al. (2008})found no significant differences in average fruit weight.

Mainly at levels above 720 g hole^-1 manure, average fruitmass for potassium treatments was higher statistically thanthat observed in treatments without potassium (Figure 2A).This response may be due to the higher nutrient contentassociated with higher levels of organic matter (^{Galvão,Salcedo & Oliveira, 2008}), and because potassium, appliedas KCl, is the main nutrient responsible for fruit quality (^{Aguyohet al., 2011}).

Contrast tests showed that an increase in fruit mass from4.2 kg^-1 (absolute control) to 7.2 kg^-1 fruit in the treatmentswith manure and potassium, indicating a 71 % averageincrease in fruit weight in fertilized soil (Figure 2B). ^{Alburquerqueet al. (2012}) found that the supply of livestock manureand mineral fertilizer significantly increased the averagemass of watermelon fruits. ^{Onyia et al. (2012}) also observedan increase in average fruit weight of cucumber asreceiving chemical and organic fertilizer.

Watermelon fruits cv. Crimson Sweet can reach usually10 to 12 kg. However, fruits below 7 kg are preferred andachieve the best price in the market, which means that thefertilization treatments applied resulted in fruits with adequateweight for the market. According to the classification proposedby ^{Carvalho (2005}), watermelon fruits obtained in treatmentswith organic and mineral fertilizer could be classifiedas «first quality» as fruit weights were between 6 and 9 kg.Except for treatments 1, 4 and 7, other treatments showed asignificant difference to the absolute control treatment (Table 4).

Table 4 Comparison between fertilized treatments and control by the Dunnet’s test (significant at P < 0.05) for average fruitweight of watermelon (kg fruit^-1). 

Productivity in treatments without potassium did not fit anymathematical model, and was thus represented by the meanvalue of 25.5 Mg ha^-1 (Figure 3A). In treatments with addedpotassium, values were adjusted to a quadratic polynomialregression model (25.071102 + 0.004496x - 0.000002^NSx²),with a maximum yield of 27.7 Mg ha^-1 corresponding to1124.0 g manure hole^-1 (Figure 3A). ^{Leão et al. (2008}) obtainedthe maximum yield of 22.5 Mg ha^-1 in watermelon fertilizedwith manure (0, 3, 6, and 9 L manure hole^-1) and NPK.^{Aguyoh et al. (2011}), working with an organic compound (0,1500, 3000, and 4500 g hole^-1) enriched with triple superphosphateand potassium nitrate, obtained a maximum yieldof 28.63 Mg ha^-1 related to the highest dose tested (4500 ghole^-1).

Figure 3 (A) Productivity of watermelon cultivated under livestock manure levels in soil without (-) and with (- - -) potassiumsupply and (B) comparison between treatments with organic or mineral fertilization (■) and control treatment (􀂅). Differentletters on top of each column mean significant Tukey’s test differences (p < 0.05) according to Tukey’s. 

The contrast tests revealed a significant increase inproductivity, from 6.8 Mg ha^-1 in the absolute control to25.7 Mg ha^-1 in treatments with manure and mineral fertilization,278 % productivity increase (Figure 3B). Yields in treatmentswith organic and mineral fertilizer is a response ofnitrogen and potassium supply to plants, which resulted inan increase in the number and weight of fruits. In other cucurbits,productivity also increases with supply of organicand mineral fertilizers to the soil, as described for cucumberplants (^{Hussein, 2009}; ^{Onyia et al., 2012}), gherkin (^{Oliveiraet al., 2009}), and pumpkin (^{Sarhan, Mohammed & Teli,2011}; ^{Santos et al., 2012}). Productivity was significantly higherin all treatments with fertilizers as compared to the controltreatment (Table 5).

Table 5 Comparison between fertilized treatments and control by the Dunnet’s test (significant at the 0.05 level) for productivityof watermelon (ton ha^-1). 

Conclusions

By raising the soil organic matter to about 3 %, the supplyof 15 kg ha^-1 potassium in Paraiba semiarid region significantlyraises productivity above the regional average.