Reuse of bottom sediment from reservoirs to cropland is a promising agroecological practice that must be rationalized

Characterization of surveyed farmers and farms

The age of the 66 interviewed farmers ranged from 20 to 75 years, with an average age of 45 years. 79% of them own cows and apply their manure in their fields. All farms except one have access to groundwater irrigation. The size of the land owned ranged from 0.2 to 6.6 hectares, with a mean of 2.1 hectares, and small and marginal farms (< 2 ha) represented 59% of the total. This was relatively larger than the average landholding in the watershed²⁸. Amongst the interviewed farmers, 31.8% own only one jeminu, while a majority (48.5%) own two, 12.1% own three and 7.6% more than three. This is comparable with the proportion found by Robert et al.²⁸ in the whole watershed for irrigable farms (farmers owning only one jeminu represent 32% of irrigable farms and 63% of non-irrigable farms). We can assume that farmers interviewed in the present study belong either to the types “Mid-size irrigated farm owners” or “irrigating smallholders” of the typology proposed by Fischer et al.³⁷ for this region.

The area of the 66 plots for which we documented sediment application ranged from 0.1 hectare to 2 hectares, with an average of 0.49 hectares. Only one of the studied plots had no irrigation facility. 85% of the studied plots were located within the Berambadi watershed boundaries (Fig. 1) and were well distributed over the watershed. Most of them belonged to the villages of Beemanbeedu (20%), Mallayanapurra (15%), Berambadi (11%) or Kallipura (9%).

Farmers’ knowledge and practice of sediment application

We asked farmers how they learned about the practice of bottom sediment application. Although the question was open-ended, all farmers gave only one answer: 66% of them answered they learned from their neighbours, 17% by themselves, 11% from family members and 6% from other farmers including some from the neighbouring state (Tamil Nadu). Surprisingly, family was far from being the main channel for knowledge transmission in our study site, although bottom sediment application is a traditional practice in South India^20,22.

We asked farmers which reasons were triggering their decision to apply sediment on a specific year. The question was open-ended, and some farmers gave more than one reason. 71% of the answers were related to a yield decrease or low yield, 13% were related to a low soil quality, 9% were related to money availability and 7% to sediment availability (empty reservoir) (Fig. 3a). Clearly, the ability to access the resource (in terms of funding or sediment availability) was not the main driver of the decision, which was rather the need to restore the land’s productivity. 10 farmers mentioned spontaneously that it was necessary to apply sediment a few years after starting irrigation on a plot (estimated from 1 to 10 years depending on the farmers) as, to their point of view, repeated irrigations decrease both soil quality and crop yield. Indeed, irrigation allows increasing cropping intensity, with up to three crop cycles per year, compared to one or two in rainfed conditions³⁷. This could lead to a decline of soil quality linked with biological degradation, nutrient depletion, soil compaction and physical degradation³⁸, and therefore a decrease of yield.

We asked farmers about the dates and frequency of sediment application. The years of applications of bottom sediment ranged from 2001 to 2022, with a peak in 2017, which can be explained as the particularly weak monsoon of 2016 left all reservoirs empty, including the Berambadi reservoir, the largest one³⁹. 41% of the farmers have applied bottom sediment only one time, on only one of their plots; 27% did it at least two times on the same plot, 18% did it at least two times but on different plots of their farmland and 14% applied bottom sediment only one time on all the plots of their jeminu (Fig. 3b). This may suggest that sediment application is essentially a one-off practice, and for the farmers who have repeated it at least twice the practice on the same plot, we found with very different time intervals between applications. While Osman et al.¹⁶ reported that “farmers believe that once applied, the impact on crop yield will remain for three years”, we were unable to highlight such a pattern from our survey.

We asked farmers their reasons for choosing the reservoir they extracted sediment from (open-ended question). 55% of the answers were related to the proximity, 39% to the sediment quality and 6% to the availability of sediment (empty reservoir) (Fig. 3c). It might suggest that logistics (including cost of transport) takes precedence over sediment composition in farmers’ decisions. Over the 66 plots, bottom sediments were extracted from 14 different reservoirs: most sediments were from Koothonur (44%), Kallipura (15%) and Deverakarre reservoirs (10.6%). The distance as the crow flies between the reservoirs and the fields where bottom sediments were applied ranged from 0.2 to 3 km with an average of 1.3 km. 45% of fields were located less than 1 km from the supply reservoir, 30% between 1 and 2 km away, and 25% between 2 and 3 km. The average distance in our study area was lower than the one observed in the state of Maharashtra by Zade et al.⁴⁰, which was 2.4 km. Hence, while motorisation probably allowed a spatial expansion of the practice, we observed that a majority of farmers still selected the closest reservoir to collect sediment.

Finally, among the farmers who applied bottom sediment at least twice (either on the same field or on different fields of their farmland), 90% of them always used bottom sediment from the same reservoir. Of the three farmers who decided to use a different reservoir, only one cited sediment quality as a reason, as his crop yield had been lower than expected after the first application. The other two were forced to change as sediment extraction was not possible in the first reservoir, either because it was full of water or because all the sediment had already been extracted.

Logistics and cost of sediment application

For bottom sediment transportation, most farmers (73%) used tractors (Figures S1c and S1d) while the others (27%) used tippers (Figures S1a and S1b). This is probably because tractors are available in the villages, while tippers belong to private contractors involved in roadworks and are less easily accessible. The average size of the plot receiving sediments was similar for tractors (0.49 ± 0.29 ha) and tippers (0.51 ± 0.39 ha). The number of loads per plot was highly variable, ranging from 9 to 400 with an average of 114 (Coefficient of Variation = 79%) for tractors and from 20 to 300, with an average of 71 (CV = 90%) for tippers. While this variability was significantly reduced by normalizing the number of loads by the field area, it remained high, with a number of loads per hectare for tractors ranging from 75 to 500, with an average of 223 loads/ha (CV = 45%), and for tippers ranging from 50 to 250, with an average of 134 loads/ha (CV = 36%) (Fig. 4a). Given these load numbers and the estimated load capacities of the tractors and tippers (around 3.5 and 11.2 m³ respectively; Supplementary Data 2), the average sediment volume applied per hectare using tractors (780 m³/ha) was around half that applied with tippers (1500 m³/ha) (Fig. 4b). This result, together with the high variability observed between plots, suggests that there is no consensus among farmers in the region on the optimum dose of sediment to apply.

These sediment volumes per hectare correspond to an average thickness of 7.8 ± 3.5 cm and 15 ± 5.5 cm for tractors and tippers respectively. Zade et al.⁴⁰ reported a similar range of magnitude in Maharashtra, where transport is mainly by tractors, with a sediment thickness of around 3 to 6 inches (7.6 to 15.2 cm).

Farmers used tippers for longer distances, with the median distance between the plot and the reservoir being 1.98 km and 0.98 km for tippers and tractors respectively. However, the cumulative distance (Fig. 4c), calculated as the product of the number of loads by the distance reservoir-field (one-way) was only slightly larger and more variable for tractors (from 7 to 565 km, with a mean of 124 km, CV = 95%) than for tippers (from 21 to 347 km, with a mean of 97 km, CV = 77%). This large variability suggests that the cumulated distance was not controlling the application rate.

The total expenditure for bottom sediment application includes only the rental of the machinery used for extraction from the reservoir, transport to the field and spreading. The sediment itself is free and no authorization is required for extraction. Most farmers paid the contractors out of their own savings (72%), while others took out a loan from other farmers (18%) or from a bank or a private company (6%). One farmer used both his savings and credit from other farmers, and one farmer did not answer the question. Finally, only one farmer benefited from a “Government” support to pay for the mechanical digger for extracting sediments from the reservoir. To explain the little use of bank loans, some farmers explained that banks would not grant loans for bottom sediment application. One farmer even admitted that a bank had given him a loan for crop cultivation, but that he had used the money for the application of bottom sediment.

The cost per plot of bottom sediment application indicated by farmers was highly variable, but on average very substantial (mean of about 90 000 INR per plot, i.e. around 1200 USD, CV = 95%). The mean cost per hectare was similar for both modes of transportation (182 000 INR/ha, i.e. around 2 460 USD, CV = 61%) compared to tippers (185 000 INR/ha, i.e. around 2 500 USD, CV 42%) (Fig. 4d). Considering the mean application rates (Fig. 4b), this means that applying a cubic metre of sediment is around twice more expensive with tractors than with tippers. It suggests that when farmers have the opportunity to access tippers, they prefer to apply more sediment than to reduce their expenses.

Farmers agreed with vehicle contractors mostly on a predetermined number of days of vehicle use (61%) or on the total cost (9%) – both equivalent to a total cost -while only 30% agreed on a predetermined number of round trips, which is equivalent to a predetermined application dose. This could explain why the cost per hectare was poorly to very poorly correlated with the number of loads (R of 0.54 and 0.28 for tractors and tippers, respectively) and the cumulative transport distance (R of 0.12 and 0.19 for tractors and tippers, respectively).

We are aware that all this quantitative information on the logistics of the practice, which calls on farmers’ memories, must be taken with caution. Surprisingly, however, farmers generally remembered the precise number of loads and the cost without hesitation, even when bottom sediment application had been carried out several years prior to the survey, whereas they had a much harder time remembering which crop was grown at the time. This could be explained by the fact that, as we pointed out above, sediment application is not a routine practice for them and, given its high cost, farmers remember its details well.

In summary, the wide variability in sediment application quantities and costs suggests that for most farmers, the objective is to apply as much sediment as possible with their available savings, in particular by choosing the closest reservoir, and not to reach a specific dose per hectare. When asked if they would have applied more sediment had it been possible, only 23% of the farmers answered that they would have put more had they have more money – and these were not the one who had put the lowest doses – while 77% answered that it was enough, confirming the lack of consensus on the objective dose.

Expected benefit of the practice and induced adaptation of the cropping system

When asked how they thought bottom sediment application affected soil quality, farmers’ most spontaneous answer was that sediment application improved crop yield (43%) – which was not providing any specific mechanism. In this case, or when no spontaneous answer was given, we suggested several options (supplementary data 1). Answers included increased soil water storage (28%), a similar benefit to manure (13%), improved soil texture and nutrient content (13%) and reduced risk of crop disease (2.6%). Aggregating these answers, we found that two third (68%) of the farmers suggested, in at least one of their answers, an impact on soil physics (soil water storage, texture and/or action similar to manure), while one third (33%) suggested an impact on soil chemistry (soil nutrients content and/or action similar to manure) and about a quarter (27%) did not provide any specific mechanism (Table 2). 77% of the farmers who mentioned an impact on soil chemistry also mentioned an impact on soil physics.

None of the farmers interviewed reported having introduced new crops after sediment application, except where this was concomitant with the installation of an irrigation system on the plot. Most of them were therefore able to compare their cropping practices for the same crops before and after sediment application (Table 2). The use of manure was unchanged for most farmers (64%), probably because almost all farmers use the manure produced by their own cows. A majority of farmers declared that they decreased their use of chemical fertilizer (58%) and the frequency of irrigation (53%). A reduction of chemical fertilizer use following application of bottom sediment was reported by several authors¹⁶, although Zade et al.⁴⁰ reported that many farmers “were reluctant to reduce their input as they felt they had invested significant money in silt application and feared loss of their investment in case of failed crops”. Decrease of irrigation frequency following sediment application is less documented in the literature.

Interestingly, farmers who did not identify a specific mechanism for the impact of sediment on soil were less likely to modify their cropping practices (Table 2). On the contrary, farmers who cited an impact on soil chemistry were significantly more likely to reduce their chemical fertilization, while farmers who cited an impact on soil physics were significantly more likely to reduce their irrigation (Table 2). This underlines the importance of helping farmers build their capacity to better understand the impact of the sediment application, so that they can better integrate it into their farming practices.

Physico-chemical characteristics of sediment and soil

The texture of the 14 sediment samples varied, according to the USDA soil textural triangle, from Sandy Clay Loam (SaClLo) to Clay (Cl) (black dots on Fig. 5a; Supplementary Table 1). While their silt contents were around 20%, the clay and sand contents were very variable. This result is in accordance with those of Braga et al.¹³ who highlighted a significant variability of particle sizes across four reservoirs in Brazil. To the contrary, surveys carried out in India seems to display a smaller variability across reservoirs within the same region, but textures varied broadly across regions: in Andhra Pradesh, sediments from 12 reservoirs belonged all to the Clay class¹⁶; in Maharashtra, sediments from 11 reservoirs belonged to the Silt Loam and Silty Clay Loam classes¹⁵; in Rajasthan, sediments from 10 reservoirs belonged to the Sandy Clay Loam, Clay Loam and Sandy Clay classes⁴¹.

Soil samples collected in the plots before bottom sediment application were also characterised by a large range of textures, from Sandy (Sa) to Clay (Cl) (red dots on Fig. 5a; Supplementary Table 1). Interestingly, although the soil samples were on average more sandy and sediment samples more clayed, their range of variation were largely overlapping (Fig. 5a). Finally, soils recently mixed with bottom sediment were characterised by a narrower range of textures, from Sandy Clay Loam (SaClLo) to Clay (Cl) (orange dots, Fig. 5a; Supplementary Table 1). For the three groups of samples (soil, sediment and soil mixed with sediment) we found a strong correlation between CEC and clay (R of 0.96, 0.97 and 0.97 for soil, sediment and soil mixed with sediment, respectively; Fig. 5b).

Chemical properties of sediments were within the range reported by Osman et al.¹⁶ in peninsular India. They were more variable than those of soils and mixtures, except for P and K content (Supplementary Figure S2). We could perform the Tukey test only on clay, silt, sand, CEC, TOC and OM, as the other parameters did not fulfil the necessary assumptions for processing an ANOVA (homogeneity of variances and normality of distributions). A significant difference between soil and sediment (p < 0.05) was observed for clay and sand (Supplementary Figure S2). Sediments had a higher clay content and a lower sand content than soils (soils mixed with sediments being intermediate, Supplementary Figure S2), as also observed by Bhanavase et al.⁴² and Canet et al.¹⁴. As a consequence, the increase of clay content due to sediment application may induce an increase of available water content (AWC) of the plowed horizon, as observed by Bhanavase et al.⁴² and Osman et al.¹⁶.

To go further, we analysed the soil property differences between soil and sediment for the 8 couples of soil and sediment collected in farmer’s plots. The non-parametric paired samples Wilcoxon test was performed on all properties as they all respect the normality of distributions assumptions. A significant difference between soil and sediment (p < 0.05) was observed for clay and sand (as previously) and also for P, CEC, TN and C/N (Fig. 6). Available P and TN content were significantly lower in sediments than in soils (Fig. 6). A significant difference between soil and sediment was also observed for CEC (Fig. 6) with higher CEC in sediment than soil. Finally, we found a significant difference between soil and sediment for the C/N ratio (Fig. 6) with higher C/N in sediment than in soil.