Forest health · method, ecology, and context
About the black ash swamp screen
What the map measures and what it only projects, why a wet forest depends on one tree, what ash loss has cost elsewhere, and why this product names black ash’s cultural importance without mapping it.
Region: Beltrami, Cass, Clearwater, Hubbard, and Itasca counties, Minnesota.
The question
A warning map, not a post-mortem
Where is black ash abundant, how soon is emerald ash borer likely to arrive, and where would losing it change the hydrology most?
Emerald ash borer (Agrilus planipennis) was found in St. Paul in 2009. Minnesota has more than a billion ash trees, more than any other state, and the Department of Natural Resources notes that many of them stand in nearly pure black ash stands on wet ground, where they help hold the water table. The Minnesota Department of Agriculture's detection record now runs to 6,609 points, and it has moved north year by year.
It has barely reached Beltrami, Cass, Clearwater, Hubbard and Itasca counties: 16 of those detections fall inside the region. That timing is the point. On the inventoried state land alone the region carries 206,164 acres of ash-bearing stands, and the counties, tribal nations, and agencies that manage them still have the option of acting before the front arrives: underplanting the species that could hold a swamp's canopy, siting biological control, and watching the wetlands most likely to change. This screen exists so those choices can be made stand by stand, with the evidence for each one on the page.
Ecological impact
One tree holds the swamp
Black ash (Fraxinus nigra) is a foundation species of the northern wet forest. In the swamps where it dominates it is often the only canopy tree, and it moves an enormous volume of water out of the ground each growing season. Experimental work at the Chippewa National Forest measured what happens when that pump stops: after simulated borer mortality and after harvest, water tables rose (Slesak et al. 2014). Remove the ash and the site becomes wetter, and the wetter it gets the harder it is for any tree to come back.
The expected end state is a change of ecosystem, not a change of species. A review of the evidence (Kolka et al. 2018) describes forested swamp converting to open sedge, cattail and shrub wetland with the loss of the ash canopy, and the Minnesota DNR describes the same conversion in its own guidance. The plants, amphibians and birds that depend on a closed wet-forest canopy lose their habitat; the flood buffering and water-quality functions of the swamp change with its hydrology.
Dominance matters more than presence. A stand that is 80% ash loses its canopy entirely; a stand that is 20% ash with cedar, black spruce and tamarack keeps one. Replacement trials in Minnesota black ash stands (Looney et al. 2015) and a species-by-species assessment against both the borer and a warming climate (Iverson et al. 2016) identify the candidates that can hold a wet site. This screen computes a replacement capacity for every inventoried stand from exactly that list, weighting each co-occurring species by how plausibly it holds the canopy: white cedar, black spruce, tamarack and red maple high; elm discounted for Dutch elm disease; aspen and pine low, because their presence marks the drier end of the gradient rather than a swamp they would hold.
Management is already under way elsewhere in the state. The Department of Agriculture's biological control program has released parasitoid wasps at 56 sites and recorded their recovery at 194 sampling points, and the underplanting trials above are the template for pre-emptive work in stands the front has not reached. The map's job is to say which stands those should be.
- Cedar, White1.0
- Spruce, Black1.0
- Tamarack1.0
- Maple, Red0.9
- Maple, Silver0.9
- Balm of Gilead0.8
- Birch, Yellow0.8
- Fir, Balsam0.7
- Hackberry0.7
- Oak, Bur0.7
- Basswood0.6
- Boxelder0.6
- Cottonwood, Eastern0.6
- Spruce, White0.6
- Willow (tree)0.5
- Birch, Paper0.4
- Elm, American0.4
- Elm, Red0.4
- Elm, Rock0.4
- Pine, White0.4
- Aspen, Largetooth0.3
- Aspen, Trembling0.3
- Birch, Heart-leaved0.3
- Maple, Sugar0.3
- Oak, Northern Red0.3
- Oak, White0.3
- Alder0.2
- Cherry, Black0.2
- Ironwood0.2
- Mountain-Ash0.2
- Oak, Northern Pin0.2
- Pine, Norway (Red)0.2
- Pine, Jack0.1
Replacement weights per species (Iverson et al. 2016; Looney et al. 2015). Ash species are zero by construction. Provisional; the first thing to ask a forester to correct.
Economic impact
Where the costs are documented, and where they are not
-
$10.7 billion
Projected cost of ash treatment, removal and replacement in urban areas of 25 states, 2009–2019
Kovacs et al. 2010, Ecological Economics
-
$1.7 billion / year
Local-government spending attributed to wood- and phloem-boring insects, with emerald ash borer the driving species, plus about $830 million a year in lost residential property value
Aukema et al. 2011, PLoS ONE
-
206,164 acres
Ash-bearing stands on inventoried state land in the five counties (9,831 stands). Federal, county, tribal and private land are not in the inventory, so the true figure is larger
MN DNR Forest Inventory, this build
The documented dollar figures for emerald ash borer are urban and municipal: street and yard trees that must be treated, removed and replaced, and the property value they carried. Those numbers are large and well established, and they are the reason most public investment so far has gone to cities.
The swamps on this map are a different economy. Black ash is a minor timber species, and the stands the screen flags are mostly remote. What is at stake there is the work the swamp does: holding water back in wet springs, filtering it on the way to lakes and rivers that carry the region's tourism and fishing economy, and keeping a forest where a marsh would otherwise be. No published study has put a dollar value on that loss for Minnesota black ash swamps, and this page does not invent one.
What the screen can do economically is direct scarce management money. Underplanting a stand before the borer arrives is far cheaper than restoring a wetland after its canopy is gone; biological-control releases and monitoring have to be sited somewhere; and a county or tribal forestry program with a fixed budget needs a defensible order in which to visit stands. The categories on the map, and the named driving factor behind each one, are built to be that order. Where a stand sits near people, the health precedent from ash loss in the eastern United States (Donovan et al. 2013) adds a second reason to prioritise it, offered as an adjacency and never as a forecast.
Cultural context
Honest adjacency
baapaagimaak
a black ash (tree), Ojibwe. Ojibwe People's Dictionary, University of Minnesota.
Black ash is the basket tree. Anishinaabe and other Great Lakes basket makers pound a black ash log along its length until the growth rings separate into splints, and weave from those. The craft is a living tradition, not a historical one, and the five counties on this map overlap the Leech Lake, Red Lake and White Earth reservations. The borer threatens the tree that tradition depends on.
This product does not map that. It maps forest health and a public invasive-species record, on a public grid, from public data. It does not map gathering places, basket-quality stands, or cultural practice, and it will not infer them. That is the same rule the traditional-foods habitat screens on this site follow under the CARE Principles for Indigenous Data Governance: publish capability and risk, withhold occurrence, and treat the nations whose land and trees these are as the authority on what is said about them.
We name the adjacency because erasing it would be its own kind of claim. This is a partner-review build. Whether and how it is useful to the tribal natural resources programs in the region is theirs to decide, and the weight tables, the categories and the copy on this page are all open to their correction.
Methodology
Three axes, one unit, declared rules
The analysis unit is a 1 km hexagon, the same 31,750-hexagon grid the traditional-foods habitat screens use, so the two products are geographically comparable. Each axis is a weighted geometric mean of its factors: one fatal factor sinks the axis rather than being averaged away, and the factor doing the most damage is the one named. A missing measurement is skipped, never scored as zero. Scoring runs in the browser from the same factor table the build used, so the evidence and threshold controls on the map recompute every hexagon on the spot.
Host
Cross-checkedTier 1, forest inventory. MN DNR per-stand species volumes on state-administered land. Ash is black, green and white ash (species 1, 39, 38); Mountain-Ash (10011) is Sorbus, not attacked by the borer, and is excluded. Ash share is ash volume over stand volume, area-weighted over the inventoried pieces of each hexagon. Covers 27% of hexagons.
Tier 2, plant-community survey. DNR Native Plant Community classes, weighted for ash: Northern Wet Ash Swamp (WFn55) and Northern Very Wet Ash Swamp (WFn64) at 1.0, communities where ash is a component at 0.2–0.5. Covers 30%.
Tier 3, wetland inventory. The updated Minnesota NWI's hardwood-wetland class, not a species map. Kept only because it ranks the inventory's ash hexagons at AUC 0.81 and the survey's at 0.85. Which tier produced a result is shown in every popup.
Exposure
ValidatedOnly 16 of 6,609 MDA detections fall in the region, so the spread model is fitted and tested statewide and then applied here.
The index is exp(−d / 20 km), with d the distance to the nearest detection, combined with MDA's own five-class introduction-risk surface for human transport. A recency-weighted form was specified and tested; on the calibration window it added 0.0005 AUC over plain distance, below the 0.01 the rule required, so it is not used.
Parameters and the front threshold (0.558) were fixed on detections through 2015 predicting 2016–2020, then tested on detections through 2020 predicting 2021–2026. Detections follow survey effort and lag infestation by years (Siegert et al. 2014): front means the detected front. Quarantine and generally-infested status are never scored; they post-date detection and would leak into the test.
Consequence
ProjectionSoils. 0.7 × the share of the hexagon in poorly or very poorly drained, hydric or organic soils + 0.3 × the Histosol share, from USDA SSURGO map units classed on the dominant component and surface horizon. Wetland. The mapped NWI wetland share. Replacement deficit. One minus replacement capacity, inventory tier only.
Absolute wetness bands were tried first and put nearly every ash hexagon in the priority class, because ash swamps are wet by definition. The bands are therefore relative: the 25th percentile of the region's 4,613 ash-bearing hexagons is the floor and the 75th the optimum. "High consequence" means the wetter, more peat-dominated, less replaceable half of the ash. It is a ranking, not a prediction, and no outcome data exist to test it.
Validation
A sensitivity figure on its own is not evidence: a model that calls the whole landscape exposed catches every new detection. Every catch rate below is reported next to the share of all units the model flagged.
| Axis | Status | Test | Result |
|---|---|---|---|
| Exposure | Validated | Temporal holdout, statewide. The model fitted on MDA detections through 2020 asked which 2 km cells would get their first detection in 2021–2026: 797 cells from 2,132 detections, against 54,801 undetected cells; 461 already-detected cells excluded. Parameters and the front threshold were fixed on an earlier window (through 2015 → 2016–2020) under a rule declared before the holdout was scored. | Rank AUC 0.861 (distance alone 0.856; MDA introduction risk alone 0.740). At the front threshold of 0.558: sensitivity 0.42 against a base rate of 0.078 (lift 5.36). The same threshold gave sensitivity 0.75 at base rate 0.031 on the calibration window: 2021–2026 arrivals jumped further from the prior record than 2016–2020 ones did. |
| Host | Cross-checked | Two DNR datasets describe ash independently: forest-inventory species volumes on state land, and the Native Plant Community survey. Compared where both cover at least half a hexagon (2,566 hexagons). Not an outcome validation. | Agreement 91% (kappa 0.36). Inventory ash share ranks survey ash-swamp hexagons at AUC 0.90; the survey ranks inventory ash hexagons at AUC 0.89. Disagreements: 180 hexagons ash by inventory only, 54 by survey only. Reported, not reconciled. The wetland-inventory hardwood tier scored AUC 0.81 against the inventory and 0.85 against the survey and was kept as the third tier. |
| Consequence | Projection | None possible. No post-mortality outcome data exist for Minnesota black ash swamps. | Not validated. The consequence axis is a modelled projection. No post-mortality outcome data exist for Minnesota black ash swamps, so nothing on this axis has been tested against an observed outcome. Factor bands are literature-based or stated assumptions. |
Specified, evaluated, dropped or deferred
Nothing leaves the model silently. Each factor that was specified and is not scored is recorded with the reason.
closed depressions deferred
Closed depressions with no outlet (lidar TWI + flow direction) need full-region lidar derivatives. native_foods/etl_lidar.py currently covers 324 of 31,750 hexagons (Bemidji smoke window) and the region-wide run is a multi-hour resumable job that has not been made. Scoring a factor that exists for 1% of units would make it decorative. Slope and relief are shown in the popup where they exist; nothing is scored on them.
drinking water proximity deferred
Surface drinking-water intakes: not built in this pass; listed as optional in the specification.
recency weighting evaluated, not used
Recency-weighted distance (tau) was specified so the active front would outweigh the 2009 metro introductions. On the calibration window (through 2015 -> 2016-2020) the best finite tau improved the rank AUC by 0.0005 over pure distance and the grid was flat from 0.9653 to 0.9658; on the holdout, pure distance 0.8562 vs nearest-5-year 0.8569. The pre-declared rule required a 0.01 gain, so tau is infinite. For the five counties this changes little: the nearest detections to every hexagon date from 2023-2026 anyway.
Health context
CDC PLACES tract estimates for coronary heart disease, COPD, current asthma and stroke are joined by the 2020 tract containing each hexagon's centroid (31,154 hexagons across 56 tracts). Donovan et al. (2013) found borer-driven ash loss associated with elevated cardiovascular and lower-respiratory mortality across 15 states, among trees near where people live. These are mostly remote swamps. The estimates are shown so partners can see where a hydrologic consequence and an existing burden coincide. They are never scored and are not a causal claim.
Limitations
- The inventory covers state-administered land only. Chippewa National Forest, county and private land fall through to the survey and wetland tiers, which see community type, not stand composition.
- The NPC ash weights and the replacement weights are provisional literature judgements awaiting partner review.
- The detected front lags the real front, and 2021–2026 arrivals jumped further than the calibrated threshold expects, so Front is conservative.
- Closed-depression detection from lidar is not built; terrain covers 324 hexagons and is shown, not scored.
- No hexagon on this map is a management recommendation without a stand visit.
Data
Sources
- Minnesota DNR, Forest Inventory: Stand Inventory Summary and Forest Inventory Species. Minnesota Geospatial Commons.
- Minnesota DNR, Native Plant Communities (Minnesota Biological Survey and DNR unit surveys). Minnesota Geospatial Commons.
- Minnesota Department of Agriculture, Emerald Ash Borer: detections, quarantines, generally infested area, introduction risk, biological control releases and recoveries. Minnesota Geospatial Commons.
- USDA NRCS, Soil Survey Geographic Database (SSURGO), via Web Soil Survey and Soil Data Access.
- Minnesota DNR, updated National Wetlands Inventory (2009–2014 imagery).
- MnGeo / USGS 3DEP, Minnesota Gen2 lidar, where noted.
- CDC PLACES: Local Data for Better Health, tract estimates.
- U.S. Census Bureau, TIGERweb boundaries.
- Ojibwe People's Dictionary, University of Minnesota: baapaagimaak.
Literature
References
- Aukema JE, Leung B, Kovacs K, Chivers C, Britton KO, Englin J, Frankel SJ, Haight RG, Holmes TP, Liebhold AM, McCullough DG, Von Holle B (2011) Economic impacts of non-native forest insects in the continental United States. PLoS ONE 6(9): e24587.
- Donovan GH, Butry DT, Michael YL, Prestemon JP, Liebhold AM, Gatziolis D, Mao MY (2013) The relationship between trees and human health: evidence from the spread of the emerald ash borer. Am J Prev Med 44(2):139–145.
- Iverson L, Knight KS, Prasad A, Herms DA, Matthews S, Peters M, Smith A, Hartzler DM, Long R, Almendinger J (2016) Potential species replacements for black ash (Fraxinus nigra) at the confluence of two threats: emerald ash borer and a changing climate. Ecosystems 19:248–270.
- Kolka RK, D'Amato AW, Wagenbrenner JW, Slesak RA, Pypker TG, Youngquist MB, Grinde AR, Palik BJ (2018) Review of ecosystem level impacts of emerald ash borer on black ash wetlands: what does the future hold? Forests 9(4):179.
- Kovacs KF, Haight RG, McCullough DG, Mercader RJ, Siegert NW, Liebhold AM (2010) Cost of potential emerald ash borer damage in U.S. communities, 2009–2019. Ecological Economics 69(3):569–578.
- Looney CE, D'Amato AW, Palik BJ, Slesak RA (2015) Overstory treatment and planting season affect survival of replacement tree species in emerald ash borer threatened Fraxinus nigra forests in Minnesota, USA. Can J For Res 45:1728–1738.
- Minnesota DNR (2003) Field Guide to the Native Plant Communities of Minnesota: The Laurentian Mixed Forest Province.
- Prasad AM, Iverson LR, Peters MP, Bossenbroek JM, Matthews SN, Sydnor TD, Schwartz MW (2010) Modeling the invasive emerald ash borer risk of spread using a spatially explicit cellular model. Landscape Ecology 25:353–369.
- Siegert NW, McCullough DG, Liebhold AM, Telewski FW (2014) Dendrochronological reconstruction of the epicentre and early spread of emerald ash borer in North America. Diversity and Distributions 20:847–858.
- Slesak RA, Lenhart CF, Brooks KN, D'Amato AW, Palik BJ (2014) Water table response to harvesting and simulated emerald ash borer mortality in black ash wetlands in Minnesota, USA. Can J For Res 44:961–968.