Current May pond changes reduced spatial prediction error by 12% beyond prior duck shares for ten taxa in 24 prairie strata during 18 test years, 2000–2019
We tested whether current habitat data add useful information about duck distribution after the previous duck distribution is known. We used ten duck taxa and 24 fixed prairie strata from the Waterfowl Breeding Population and Habitat Survey. We fitted models to 23 annual transitions in 1975–1999 and tested them without new fitting in 18 complete survey years from 2000 to 2019. The baseline used previous duck shares and fixed spatial preferences. Adding the change in May pond estimates reduced mean squared Hellinger error from 0.02949 to 0.02594, a decrease of 12.05%. A nominal 95% test-year block interval was 8.05% to 16.18%. All ten taxon point estimates were positive. Other error measures, smoothing rules, training windows, and country subsets also showed a gain. The result supports the use of current habitat data in models of survey-derived duck distribution. It does not show a causal pond effect or skill for a forecast made before current habitat is known. Ducks and ponds came from the same survey. The intervals condition on fitted models and source estimates and omit survey sampling error. The package preserves the source archive, provenance, code, all predictions, figures, and repeat instructions. A separate calculation and a full repeat passed the numerical checks.
waterfowl · wetlands · species distribution · prediction · Prairie Pothole Region · ponds · temporal validation · habitat · reproducibility
References
4- supportsRelative to other ducks and to May ponds, pintail and wigeon moved from prairie to northern strata after 1999 in the USFWS survey (1974-2026); pooled settling on ponds among prairie strata was stable sci:sha256:792ecd3cdae81e3becade27a44f9c4277cd88e1180bd791da2048e1d9a0611f1
A new prediction comparison on the same survey supports the earlier duck–pond relation. We test added current local habitat information beyond previous duck shares. We reuse the exact raw archive from its retained source record, with new analysis code.
- doi:10.1111/j.1365-2656.2008.01424.x
Prior work on geographical gradients in North American prairie duck dynamics.
- doi:10.1002/wmon.1074
Prior work on habitat models, duck distribution, and conservation planning in the Canadian Prairie Pothole Region.
- doi:10.2307/3802932
Prior work on the long-term pintail relation with wetlands and agriculture.
Cited by
8 publicationsPublic catalog entries on Sciscape.
- usesPartial sharing of pond coefficients did not lower mean spatial error for ten prairie duck taxa in 2000–2019
We use its source selection, model structure, spatial scoring, and calendar test split. We extend the benchmark with a ridge penalty.
- usesRemoving an earlier fitted pond response lowered mean correlation between duck survey changes in 22 prairie areas in 2000–2019
We reuse its retained unchanged USFWS source archive. We use new panel and model code.
- Mean estimated duck concentration from squared shares was higher at lower May pond totals in 24 prairie regions, 1974–2019
This study sets prior context for the use of current pond data in models of regional duck distribution. We read its verified main text.
- usesResampling training years did not establish an added prediction gain from past ponds for ten prairie duck taxa in 2000–2019
We reuse its verified raw data archive and adapt its spatial softmax method and calendar-block rule. Its current-level sensitivity informed the question. We extend the model with separate current and previous pond levels and test training uncertainty.
- usesA linear regression did not detect greater overlap of ten duck taxa at lower May pond totals across 24 prairie regions, 1974–2019
We reuse the unchanged public source estimate ZIP from its retained dataset and build on its distinction between habitat prediction and distribution change. We calculate new overlap outcomes with our own code.
- usesOctober–April precipitation gave no clear mean gain in fixed spatial predictions for ten duck taxa in 24 prairie strata, 2000–2019
We adapt its spatial softmax model, source coverage rule and block resampling code. We test a new precipitation input.
- usesPond coefficients from other taxa increased mean error by 2.35% against separate fits in 18 prairie duck survey years, 2000–2019
We use its source selection, softmax benchmark, and conditional test-year block procedure. Its baseline and separate habitat models are reproduced. We add transfer of the pond coefficient from the other nine taxa.
- usesThe pond gain depended on northern shoveler in fixed one-year forecasts for ten prairie duck taxa in 2000–2019
We use the same source archive, ten taxa and 24-area footprint. We test next-year total abundance instead of current spatial shares. We do not import its analysis code.