Sandtrout Membrane Ultrastructure and Active Water Uptake: Electron Microscopy and Field Core Trials in the Southern Erg Margin, 10236–10238 AG

1. Introduction

Sandtrout, the little makers, are the vector stage of Shai-Hulud. They are small, leathery organisms of the sand column below the zone of daily re-evaporation, and they encapsulate free water wherever they find it. The Imperial Planetologist Pardot Kynes and his son Liet-Kynes, whose survey registers run from the 10150s AG to 10190 AG, identified this behaviour as the reason Arrakis stays dry (Office of the Imperial Planetologist, Arrakis, 10152–10190 AG). The Fremen greening programme later took the finding into its planning. That programme still has to deliver collector water faster than the sandtrout remove it.

What the survey recorded was the outcome of sequestration, never its mechanism. A membrane that only absorbs water down an osmotic gradient would be limited by the water potential of the surrounding sand. It could not take water from pore water saltier than its own fluids. A membrane that pumps water, by contrast, could keep taking it up at low moisture, but only at a metabolic price. The distinction sets the upper limit on how much water a population of little makers can remove. It therefore bears on any water budget for the planet and on the pace of greening (Straka, 10229 AG).

We apply a three-step test. First we ask whether measured uptake exceeds what passive diffusion allows under any of its plausible parameter settings. Second we ask whether an active pump would be energetically affordable for an organism of sandtrout size. Third we scale the measured rate to the planetary population and compare it with the water that actually enters the sand today. Arrakis has no rain. Its present influx is dew and atmospheric moisture, the same sources that windtraps and dew precipitators capture.

Our vantage is about 10240 AG, within a decade of the fieldwork. We use only observations made under survey protocol and treat earlier accounts as background. The study concerns uptake across the membrane. The subsequent sealing of water into pockets, which is what the Kynes account calls sequestration, lies outside what our trials observed.

2. Methods

Specimens and fieldwork. Between 10236 and 10238 AG we collected 34 sandtrout from eight aggregation sites in the southern erg margin, with the consent of the cooperating sietch councils. Mean wet mass was 38 g (SD 11) and mean membrane area 41 cm² (SD 9). Peripheral membrane was fixed in the field and sectioned for transmission electron microscopy (TEM) and scanning electron microscopy (SEM). A preliminary morphometric account of the first specimens has appeared separately (Threll, 10236 AG). As a passive-absorption comparison we prepared root epidermis from 12 poverty grass specimens taken from Survey planting stations. Pores were counted in ten imaged fields of 0.25 µm² per specimen, giving 340 sandtrout fields and 120 poverty grass fields.

Uptake trials. Twelve monitoring sites were chosen in sand occupied by sandtrout, eight of them at the collection sites. Ambient moisture at 0.8–1.2 m depth, measured before the trials, averaged 0.3% by volume (range 0.1–0.6%). At each site we buried two 10-litre cores of local sand, wetted to 4% volumetric water content (400 mL), at 0.8–1.2 m and about 1 m apart. One core sat in an open sleeve that sandtrout could enter. The other sat in a fine-mesh sleeve that excluded them but allowed vapour exchange and drainage. The contrast of interest was sandtrout access versus exclusion on identically wetted sand.

Each week for eight weeks both cores were lifted and emptied through a 5 mm sieve over a closed tray. Sandtrout retained on the sieve of the open core were brushed free of sand over the tray, counted and held in a sealed dry container. The sand of each core was then returned to its sleeve and weighed to 0.02 g, re-wetted to 400 mL and reburied, and the animals were released into the open sleeve. Because the animals were removed before weighing, the open-core loss includes the water they had taken up and carried out of the core in their bodies. Net sandtrout uptake for a core-week was the open-core loss minus the exclusion-core loss. Twelve sites over eight weeks gave 96 core-weeks. Measurement error was assessed in two ways. Repeat weighings of the same core differed by at most 0.02 g, and handling blanks, in which cores without sandtrout were sieved and refilled, lost 0.03 g per cycle (SD 0.02) in both sleeve types. Contact trials put water adhering to brushed animals at under 2% of the weekly net loss.

Per-individual rate. Each net weekly uptake was divided by seven and by the number of sandtrout counted in that open core. Mean occupancy was 20.2 individuals per core (SD 5.1), far above the ambient density, because sandtrout collect at wet sand. The resulting rates are therefore uptake capacities under water-rich conditions. The per-individual rate, a continuous outcome with weeks nested within sites, was analysed with a linear mixed model with a random intercept for site. The 95% confidence interval for the mean was taken from that model with 11 degrees of freedom.

Passive benchmark. The passive expectation came from an osmotic-gradient uptake model parameterised with membrane permeabilities from eight arid-adapted taxa of Arrakis and neighbouring Imperial worlds, including poverty grass and the dune-fixing shrubs of the Survey stations. These were compiled from 23 field and laboratory studies (Ossery, 10226 AG). Scaled to the sandtrout membrane area and to the water content of the wetted cores, the model gives 0.3–0.8 µL per individual per day across all variants of its gradient assumptions. Our primary comparison is against the most permissive variant, 0.8 µL per day. As a secondary summary we treated the passive value as uniform on 0.3–0.8 and derived the distribution of the ratio of observed to passive uptake by Monte Carlo simulation, drawing the observed mean from its mixed-model sampling distribution.

Energetics and scaling. Minimum osmotic work was computed as the product of pore-water osmotic pressure and the volume taken up. A pump efficiency of 10% (range 5–30%) was then applied. Resting metabolism for a 38 g desert ectotherm was taken from the mass-scaling relation of Veradin (10218 AG). Population size came from excavation transects in the open erg (Tabrizan, 10231 AG). Present-day influx was taken from the Kynes-era buried lysimeter and dew-gauge registers (Office of the Imperial Planetologist, Arrakis, 10150–10190 AG), whose gauges had a detection limit of about 3 × 10⁻⁵ mm per year.

3. Results

Peripheral membrane thickness averaged 12.4 µm (SD 2.1; range 8.2–17.3; n = 34). Both TEM and SEM showed a dense, regular array of pores, with a mean density of 8.2 per µm² (SD 1.3) and a mean diameter of 2.3 nm (SD 0.4). Poverty grass root epidermis carried 0.6 pores per µm² (SD 0.3), with a mean diameter of 9.1 nm (SD 3.2). In a negative binomial mixed model of pore counts per field, with field area as offset and a random intercept for specimen, sandtrout membrane had 13.7 times the pore density of the comparison tissue (95% CI 11.2–16.8; p < .001). Pore diameter was also about half as variable in the sandtrout, with coefficients of variation of 0.17 and 0.35 respectively.

Exclusion cores lost 1.9 g per week (SD 0.5) to drainage and vapour exchange. Because paired cores lay side by side, their losses moved together, and open cores lost an additional 0.34 g (0.34 mL) per week (SD 0.08), or 34 µL per litre of core per week. Dividing that net loss by seven and by the mean occupancy of 20.2 gives about 2.4 µL per individual per day. The mixed model estimated a mean of 2.4 µL per individual per day (95% CI 2.1–2.7). The between-site standard deviation was 0.50 and the residual standard deviation 0.35, so sites accounted for about two-thirds of the variance. The raw standard deviation across the 96 core-weeks was 0.6.

Against the most permissive passive variant of 0.8 µL per day, observed uptake was 3.0 times higher, and 2.6 times higher at the lower confidence bound of 2.1. Every variant of the passive model is therefore exceeded by a factor of at least about 2.6. Against the uniform-averaged passive expectation the median ratio was 4.4 (95% interval 2.9–7.8), and no simulated draw fell below parity. At mean occupancy, the most permissive passive variant would remove about 0.11 g from an open core per week. The observed net loss exceeds that by about 0.23 g, roughly six times the combined weighing and handling uncertainty of about 0.04 g.

Pore water in the wetted cores carried about 3.2% dissolved salts by mass, an osmotic pressure of about 2.5 MPa. Taking up 2.4 µL (2.4 × 10⁻⁹ m³) against that pressure requires at least 2.5 × 10⁶ Pa × 2.4 × 10⁻⁹ m³, or 6.0 × 10⁻³ J per day. At 10% pump efficiency the cost is 0.06 J per day (0.02–0.12 J across the efficiency range). The scaling relation gives a resting metabolism of about 1.3 kJ per day (plausible range 0.7–2.5 kJ). The pump would therefore consume about 0.005% of the energy budget, and under 0.02% at the least favourable bounds.

4. Discussion

Two independent lines of evidence point to active transport. The membrane's pores are much finer, denser and more uniform than those of a passive absorptive tissue, which fits a structure built to move water selectively. Field uptake also exceeds the most permissive passive variant threefold, and by at least about 2.6 times at the lower confidence bound. Energetics show only that nothing prevents a pump from existing, since its cost is a tiny fraction of resting metabolism. Dry sand would raise that cost, because matric suction adds to the osmotic term. Even a suction of 50 MPa, however, at the lowest efficiency and lowest metabolism considered, keeps the pump below 0.4% of metabolism.

Planetary scaling needs more care. Transects give a mean of 3.3 individuals per m² of occupied erg (range 1.1–8.9). Using the Survey's working figure of about 3 × 10¹⁴ m² of open desert, with roughly 30% of it occupied, the population is near 3 × 10¹⁴ (range 1–8 × 10¹⁴). Each individual can take up 2.4 µL per day, or 0.88 mL per year, so the population capacity is about 2.6 × 10¹¹ litres per year (range 8.8 × 10¹⁰ to 7.0 × 10¹¹). The lysimeter registers put the dew-derived water that passes below the zone of daily re-evaporation at about 1.5 × 10⁻⁴ mm per year, with readings from below detection to 6 × 10⁻⁴ mm.

Which area the influx is taken over matters. Sandtrout can intercept only the water that reaches ground they occupy, so the direct comparison is with influx into the occupied 30%, about 1.35 × 10¹⁰ litres per year (upper bound 5.4 × 10¹⁰). Capacity exceeds that by a factor of about 19, and by at least about 1.6 across the combined ranges. A stricter test asks whether the population could absorb the influx over the whole open desert, about 4.5 × 10¹⁰ litres per year (upper bound 1.8 × 10¹¹). There the factor is about six, with a lower end near 0.5. Because the lowest readings fall below detection, neither ratio has a finite upper bound; at the detection limit the whole-desert factor would be about 80.

The survey's claim that sandtrout keep Arrakis dry thus concerns two quantities. The first is the stock the little makers sealed away as the planet dried, which our data do not measure. The second is the present flux. Our measurements show only that the population can plausibly intercept the small quantity of water that now reaches their depth. For the greening programme the local picture matters more than the planetary one. Collector water applied to a planting is many times the natural infiltration, and trajectory models of windtrap deployment already treat sandtrout uptake as a loss competing with plantings for delivered water.

A time-to-event analysis of archived aggregation sequences found that the interval from aggregation to spice blow shortens as aggregate mass grows, and proposed that a colony must accumulate a threshold quantity of water at a rate set by its number of individuals. Our per-individual rate supplies the quantity that such an account requires, although these data cannot test it. Whether a given aggregation proceeds to a pre-spice mass also varies with site soil geochemistry, which our trials did not vary.

5. Limitations

The uptake rate is a capacity measured on wetted sand that drew sandtrout in at high density. Realised uptake in dry ambient sand is likely lower, and the planetary figure should be read as an upper estimate of what the population could remove. The per-individual denominator is the count at each weekly lift; animals that left the core during the week carried water out without being counted, which biases the rate by an amount we could not determine. Our cores sat at 0.8–1.2 m. The Kynes-era registers place sealed pockets most often between about 1 and 4 m, so our cores sampled only the upper edge of that range. All sites lay in the southern erg margin, and the deep desert may differ.

Our passive benchmark rests on permeabilities from other taxa scaled to sandtrout geometry, and a single comparison tissue was imaged. Population size and deep infiltration are each uncertain by close to an order of magnitude, and the lysimeter registers are fifty to ninety years old. We therefore report the budget comparison as a range and do not claim a single decisive factor.