Flowering-Window Salinity and the Water Balance of Pundi Rice Paddies in Three Caladan Estuaries: Tidal Gate Rules, Leakage and Block Yields, 10231–10237 AG

1. Introduction

Across the 82 block-seasons of pundi rice reported here, the paddy that received the most water, 1,307 mm between transplanting and harvest, yielded 6.6 t/ha, and the one that received the least, 889 mm, yielded 6.7 t/ha. Supply varied by 47% from least to most. The poorest crop of all, at 2.3 t/ha, was not the driest paddy. It was the saltiest, at 9.1 dS/m in the thirty days around flowering.

We write c. 10238 AG, from the coast of the ocean world that House Atreides held for twenty-six generations before Arrakis. Pundi rice is Caladan's staple crop, grown in flooded fields on diked blocks called polders at the heads of short tidal estuaries. Freshwater is not scarce there. The hazard is salt, which the tide carries up the channel that supplies the blocks, in amounts set by the fortnightly spring–neap cycle and by how hard the river pushes back (Polgarrow, 10227 AG; Trelisk, 10234 AG).

The study began as an inversion of dry-side practice. The third author trained on Arrakis, where the discipline is to close the root-zone water ledger and count every litre (Pask, 10226 AG). We applied her protocol on Caladan to ask whether a ledger that closes also explains the crop, measuring salinity in two windows because paddy grasses are most salt-sensitive at establishment and at flowering (Nansledan, 10221 AG). We also asked whether keepers' gate rules change the answer, and how much salt enters through shut gates.

2. Methods

Fieldwork covered seven growing seasons, 10231–10237 AG, on twelve blocks totalling 96.6 ha in three estuaries of the northern littoral, Porthel, Lanreth and Carrow, four blocks each. Block boundaries, gate positions and sill levels come from the household polder survey (Atreides Household Estate Office, 10152 AG); a sill is the threshold over which a sluice gate admits channel water. In 10234 AG a surge breached the seaward wall at Carrow and flooded blocks CA-3 and CA-4, and both crops were lost, leaving 82 of 84 block-seasons, a block-season being one block in one growing season.

Gates admit water on the rising tide, while the channel stands above the sill, and blocks drain by gravity on the ebb. Under the every-tide rule (three blocks) the keeper opens the gate on each rising tide and shuts it at high water. The spring-closure rule (three blocks) adds a shutdown around each spring-tide high. The conductivity-gated rule (six blocks) admits water only while intake conductivity stays below 1.5 dS/m. Of 19 surviving household instructions to gate-keepers, 13 prescribe the every-tide rule, 5 add spring closure and one leaves admission to the keeper's judgement; none mentions the crop's stage (Atreides Household Estate Office, 10118–10190 AG). The conductivity rule postdates that series: the six blocks that use it belong to two cooperatives that adopted a meter-based rule after 10220 AG. Loggers show the declared rule followed on 93% of 11,152 block-days.

Seedlings were transplanted about 30 days after sowing, and the season runs 136 days to harvest. The establishment window is days 1–21 after transplanting; the flowering window is days 64–93, from panicle initiation to the end of flowering on weekly phenology plots. Yield came from six 10 m² crop cuts per block-season, as rough rice at 14% moisture.

The ledger followed the third author's closure protocol (Pask, 10226 AG): rain at one gauge per estuary, intake from gate openings and a discharge rating, drainage at outfall weirs, evapotranspiration from canopy tanks, percolation from sealed plots, storage from staff gauges. The closure residual is inflow minus outflow minus storage change. Leakage, the water crossing a shut gate while the channel stands above the sill, came from 36 closed-gate seepage tests at spring-tide high water (mean 1.9 L/s, range 0.8–3.6), applied to the logged hours each gate stood shut under a high channel.

Root-zone salinity is the conductivity of porewater at 15 cm from five fixed samplers per block, read weekly and averaged within each window, in dS/m. Intake loggers gave channel conductivity, and salt imported is depth times conductivity times 6.4 kg per hectare per millimetre and dS/m, for admitted water and leakage separately.

Yield is continuous, so we fitted linear models with a fixed effect for each block, which absorbs soil, elevation and gate position, standard errors clustered by block (12 clusters) and intervals from a t distribution on 11 degrees of freedom. Predictors were conductivity in each window and supply (rain plus intake plus leakage) per 100 mm; a second version added season effects. To locate a bend in the flowering-window relationship we fitted separate slopes either side of a break point, searched from 1.0 to 4.0 dS/m in steps of 0.1 and kept the break with least residual error. Because the data chose it, p-values for the slopes are conditional on it, and we give a likelihood-based range for its position (Tavistrane, 10229 AG). The driest season was compared with each block's other seasons by 12 paired differences.

3. Results

Water supply averaged 1,097 mm per block-season (range 889–1,307 mm): rain 611 mm, intake 467 mm and leakage 19 mm (Table 1). Evapotranspiration and percolation together took 716 mm, so every block-season supplied at least 163 mm, and at least 21%, more than it lost. Drainage carried off 382 mm, about 35% of inflow. The ledger closed to −3 mm on average (−0.3% of inflow, SD 2.2%); 1 block-season exceeded 5%, at 7.7%.

Supply does not sort the crops. In 10233 AG blocks CA-2 and CA-3 at Carrow received 1,026 and 1,025 mm and yielded 3.5 and 6.0 t/ha, with flowering-window conductivities of 5.9 and 3.0 dS/m. Within blocks, supply alone is weakly and positively related to yield (Table 2, model A), but that is the rain's doing: each extra 100 mm of supply lowered flowering-window conductivity by 0.76 dS/m (95% CI −1.28 to −0.24, p = .008), because wet seasons dilute the channel. Flowering-window conductivity alone explains 68% of within-block variation (model B). With both windows and supply together (C) its slope was −0.52 t/ha per dS/m (95% CI −0.68 to −0.36, p < .001), while the coefficients for supply and establishment-window conductivity were indistinguishable from zero; season effects (D) left the slope at −0.57. The windows are correlated (r = 0.84), so an establishment effect of up to about 0.2 t/ha per dS/m cannot be excluded.

The relationship bends. The best-fitting break point was 2.1 dS/m (approximate 95% range 1.8–2.7). Above it yield fell by 0.57 t/ha per dS/m (95% CI −0.76 to −0.39, p < .001, conditional on the selected break); below it the slope was −0.07 (−0.39 to 0.25; model E, within-block R² 73%). Conductivity exceeded 2.1 dS/m in 22 of 82 block-seasons (27%), with mean yield 5.52 t/ha against 6.65 t/ha in the other 60.

In the driest season, 10233 AG, rain averaged 477 mm, 22% below the seven-season mean, and supply fell by 123 mm within blocks (95% CI −151 to −95), yet keepers admitted 32 mm more intake (95% CI 6 to 59, p = .022). Flowering-window conductivity rose by 1.9 dS/m (95% CI 1.0 to 2.8, p < .001), the pattern expected of a salt wedge pushed upstream by weak river flow, and yield fell by 0.89 t/ha (95% CI −1.55 to −0.23, p = .013). That season 9 of 12 blocks exceeded 2.1 dS/m, against 13 of 70 block-seasons in the other six.

Gate rules separated the blocks (Table 3). Conductivity and yield ordered every-tide, spring-closure and conductivity-gated blocks from worst to best, although every-tide blocks received the most water, and the three conductivity-gated exceedances of the break were all in 10233 AG. Holding every exceedance at 2.1 dS/m would, by the slope above the break, have raised yield by 0.64 t/ha (95% CI 0.43–0.85) on every-tide blocks, 0.21 (0.14–0.28) on spring-closure blocks, 0.03 (0.02–0.04) on conductivity-gated blocks and 0.23 (0.16–0.31) overall, conditional on the selected break and the exceedances observed.

Shut gates were not tight. Leakage averaged 19 mm per block-season, 4.1% of intake volume, but carried water at 4.5 dS/m (range 1.9–10.4) against 1.4 dS/m (0.6–5.1) for admitted water in the flowering window, and so brought 11.2% of the salt imported. The share rose as rules tightened (Table 3), and at the mean seepage rate the three kinds of block stood shut under a high channel for about 160, 200 and 270 hours a season.

4. Discussion

What limited the crop was salt in a thirty-day window, not the amount of water. The volume terms of a ledger with a mean closure residual of 0.3% of inflow added nothing detectable once salinity was in the model, and every block-season held a volume margin of at least 21%. The dry side makes the contrast plain: in the companion reserve model for the Fremen sietches the operating margin is itself a volume, a median of 31% of annual requirement under full compliance, and discipline decides whether it stays out of the high-risk band. Here the scarce thing was a tidal window, the part of the rising tide in which the channel is fresh enough to admit.

Gate rules matter through that window. Conductivity gating kept flowering-window salinity under the break in 37 of 40 block-seasons, and its three failures all fell in the one drought season. The every-tide rule, the one most household instructions prescribe, secures supply but imported the most salt, 7.6 t/ha a season. None of the three rules is keyed to the crop's stage, though only the thirty days at flowering showed a detectable cost; a rule reserving the cleanest tides for that window is the obvious next trial, which our data do not test.

In conductivity-gated blocks 17.7% of imported salt arrived through shut gates; if conductivity scales with salt imported, that is a first-order 0.3 dS/m of the 1.53 dS/m mean. That is too small to change the average crop but it matters in a drought: in the three conductivity-gated exceedances of 10233 AG, removing leakage salt in proportion would have lowered conductivity by about 0.4 to 0.9 dS/m, enough to bring two of the three to or under 2.1 dS/m.

The rules that protect the crop best also shut gates when the estuary is busiest. Earlier work by one of us places most upstream passage of juvenile fish at these sluices within two days of spring-series high waters (Tremmere, 10233 AG), the very tides that spring closure excludes and that conductivity gating excludes whenever they are salty; we did not measure passage here. What the data support is narrower than a prescription. Across 82 block-seasons on three estuaries, yield followed root-zone salinity in the thirty days around flowering, with a break near 2.1 dS/m; supply added nothing detectable once salinity was counted; and shut gates leaked a small volume of salty water whose share of the salt imported grew as the rules tightened.

5. Limitations

This is an observational design. Gate rules were inherited, not assigned, and with three, three and six blocks per rule the contrasts rest on few units. Block effects remove fixed differences, but a rule could travel with an unmeasured trait that varies by season. Seven seasons contribute, one a drought, and all blocks share each season's weather; standard errors clustered on 12 blocks can be too small.

The surge removed two block-seasons from the analysis, so the slope above 2.1 dS/m rests on conductivities up to 9.1 dS/m and says nothing about flooding by sea water. Conductivity from five samplers per block is a noisy measure of the root zone, which would bias slopes toward zero.

River discharge was not gauged, so the salt-wedge account of 10233 AG is inferred, not measured. Leakage depth is an estimate from 36 tests, a mean rate and logged hours. Everything here concerns one crop, one coast and transplanted paddies; we draw no conclusion for estuaries with large rivers, other tidal regimes, or seasons wetter or drier than the seven observed.