I crouched on the slope and noticed one small stem at eye level.

It was a konara oak seedling. About the length of my finger. And when I looked at what it was growing in, there was no soil — not yet. Just a layer of dry, recently fallen leaves. Leaves that crinkle when you move them. Leaves that won't become humus for several more years.

But inside that leaf litter, the seedling had pushed its thin roots down and was standing.

This is Hounji Temple in Kameyama City, Mie Prefecture, Japan. On March 27, 2026, we installed a shigara — a traditional Japanese woven-stake erosion control structure — on this bamboo-covered hillside. That was 75 days ago.

When I came back on June 13, I knelt on the slope and started counting.

31 seedlings.

What Is Shigara?

Shigara doesn't translate directly. The closest English equivalent is "brush check dam" — but that misses something important.

A shigara is designed to disappear.

We drive bamboo stakes into the slope along the contour lines, weave cut branches between them, and pack the sheltered space behind with fallen leaves and branch material. There's no concrete. No heavy machinery. No permanent structure.

The logic: slow the water down. Give the soil time to settle. And in that time, let plants establish their roots. When the bamboo stakes eventually rot — in 5 to 10 years — the roots should take over the work the stakes were doing.

The structure disappears. The forest continues.

Think of it like Beaver Dam Analogs (BDAs) used in North American stream restoration: semi-permeable barriers placed to disrupt erosive flow without stopping it entirely. But shigara goes further — we're not just stabilizing soil. We're rebuilding a forest.

For this installation at Hounji Temple, the numbers were: 29 stakes, 80 crosspieces, 100 kg of brush material, and 120 kg of konara oak leaf litter — with acorns mixed in.

The Problem: Four Barriers

Here's what makes this hillside unusual. On a normal bamboo-cleared slope, konara oak doesn't just come back on its own. In fact, it almost never does.

There are four barriers, stacked on top of each other.

First: konara acorns don't stay dormant in the soil. They germinate within 1–2 weeks of falling and die quickly if they dry out. So even after clearing the bamboo, there are no seeds waiting in the ground.

Second: the animals that disperse acorns — jays, squirrels, field mice — won't enter a dark, food-poor bamboo stand. No animals, no seeds arriving.

Third: bamboo regenerates fast. Even after clearing, it resprouts aggressively. A konara seedling can't grow fast enough to outcompete it before the canopy closes again.

Fourth: konara oak depends on ectomycorrhizal fungi (ECM) to establish its root system. But when the adult oaks disappeared from this site, the ECM fungi disappeared too. A seedling can sprout — but without its fungal partner, it can't root.

Our design was built to address all four barriers at once. The leaf litter came from nearby konara woodland, carrying dormant fungal spores. The acorns provided seeds directly. The shigara structure created shade and moisture control. Ongoing bamboo thinning gave the seedlings a light window.

The result: 31 seedlings, 75 days in.

What Came Up

Here's the full plant count from the June 13 survey:

Konara oak (Quercus serrata): 22. Muku tree (Aphananthe aspera): 2. Kuma-shide or Yashya-bushi: 2. Kunugi or Abemaki oak: 1. Tsuruume-modoki vine: 1 (possible). Unidentified: 3. Total woody seedlings: 31.

Direct emergence from brush material (soda): cherry seedling: 1. Unidentified: 2. Total: 3.

That last category is worth noting. Three seedlings appear to have sprouted directly from the cut branch material used in the structure.

This is called a "live stake effect" (lebendverbau in European bioengineering) — when branch material is still viable and roots directly into the soil. We used locally harvested material from the same site, which may have contributed.

Three Ways to Read a Weed

When we arrived in June, the shigara had a mat of low vegetation growing over it. Weeds — but not all the same. We don't pull everything. We don't leave everything. We sort each plant into one of three categories.

Pull now: Yabu-myoga (Pollia japonica) spreads by underground rhizome and directly displaces oak seedlings. Yabu-garashi (Cayratia japonica) climbs and strangles young trees. These two we remove by hand, one plant at a time. (Note: the area fertilized with rice bran showed particularly dense yabu-myoga growth — we'll reduce or eliminate bran on the next installation.)

Wait: Tsuyukusa (Commelina communis) and Inutade (Polygonum longisetum) are annual pioneer species. They're holding the disturbed surface while slower plants catch up. They'll self-limit as the soil stabilizes. We leave them.

Keep: Inokodsuchi (Achyranthes bidentata) tolerates shade and coexists with seedlings. Kusa-ichigo (Rubus hirsutus) attracts birds with its fruit — and birds bring seeds of other species. These are assets. We protect them.

Pull, wait, keep. That sorting is what separates hand-weeding from forest restoration.

Fragile and Resilient Are the Same Thing

I want to return to that first seedling — growing in dry leaves with almost no soil around it.

Konara acorns store nutrients inside the seed itself. That reserve is enough to push out the first root and first leaves, even with no humus. But it runs out. If the roots don't reach mineral soil before the reserves are gone, the seedling dies.

This is why most seedlings don't survive. Even in a healthy forest, most acorns produce seedlings that die within weeks. The ones that make it are the ones that happen to land in exactly the right spot.

When I pulled back the leaf litter around that first seedling, the underside was damp and alive. Pillbugs, earthworms, earwigs — all inside the shigara structure. Decomposition had already started.

Fragile and resilient are not opposites. They are two faces of the same thing. Fragile doesn't mean weak. Standing while carrying fragility — that is what strength looks like.

We kneel on that slope and pull the invasive plants away from those seedlings, one at a time, because we want their roots to reach mineral soil before their reserves run out.

The Timeline: 1 Year, 5 Years, 10 Years

Year 1 (now): Herbaceous plants establish. Fungal networks begin forming underground.

Year 5: Low shrubs and konara seedlings gain height. The slope has texture and root structure again.

Year 10: The bamboo stakes begin to rot. By then, the plant roots should hold the slope on their own. The structure disappears. The living system continues.

Disappearing is the design. That's not failure — that's completion.

We build with the understanding that our role is to be temporary. The structure is a scaffold, not a solution. The forest is the solution.

A Note on Biodiversity

The Hounji Temple bamboo woodland was designated as a Kameyama City Biodiversity Coexistence Area in October 2025.

Reducing disaster risk and calling wildlife back — those are the same work. The trail camera near the shigara has already recorded a tanuki (raccoon dog). Tanuki use fixed latrines, and their scat is one of the primary vectors for konara seed dispersal in this region. We didn't plant that animal. We made conditions where it was worth returning.

Still Watching

31 seedlings, 75 days in. The slope is still fragile. Some of those seedlings may not make it to the soil.

The one I keep thinking about is that first one — thin stem, dry leaves, no soil yet, roots reaching down anyway. I still don't know if it will make it. That uncertainty is the honest answer.

But I came back to kneel on that slope and pull the invasive plants away from it, one plant at a time.

That's how this kind of work goes: you set something in motion, and then you pay attention to what it does next.

We'll come back in the autumn.