Climate Change Contributes to Loss of Wild Rice (Manoomin/Psiŋ), Threatening Indigenous Lifeways

doi:10.21203/rs.3.rs-4659778/v1

Download PDF

Analysis

Climate Change Contributes to Loss of Wild Rice (Manoomin/Psiŋ), Threatening Indigenous Lifeways

https://doi.org/10.21203/rs.3.rs-4659778/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Climate change threatens lifeways of Indigenous Peoples, impacting their rights to self-determination and sovereignty. In the Laurentian Great Lakes region, Indigenous communities have experienced declines of wild rice (Ojibwemowin: Manoomin; Dakodiapi: Psiŋ; Latin: Zizania palustris), a sacred plant central to their culture. Past studies have implicated climate change in these declines but without quantification. Leveraging long-term, large-scale monitoring data collected by tribal organizations, we analyzed wild rice density and harvest data in relation to key climate conditions expected to impair wild rice’s vulnerable early life stages. Our results indicate that wild rice available for tribal harvest off-reservations has declined regionally by ~ 5–7% annually and likely will continue in part due to anthropogenic climate change, specifically increased early summer precipitation and warmer winters with less snow and shorter ice-duration. This decline has infringed on Indigenous lifeways by reducing off-reservation tribal harvest, a right guaranteed by treaties with the US government.

Climate change disproportionately disrupts Indigenous Peoples’ land-based lifeways^1,2, which are legally protected by their inherent rights to self-determination³ and sovereignty⁴. Despite this outsized impact and direct consequence for their survival, Indigenous Peoples have been largely excluded from mainstream climate research⁵. For the Indigenous Peoples of the upper Laurentian Great Lakes, potential climate change impacts on their sacred food^6,7, wild rice (Zizania palustris L.), are of concern^8–10 as this region experiences warming exceeding average global rates¹¹. For the Ojibwe, one of three Peoples among the Anishinaabeg, wild rice was integral in the prophecy guiding their migration to this region¹², and now serves as a cultural foundation protected for harvest under treaties signed with the United States^6,13,14. While wild rice is also culturally important for the Dakota, who call it Psiŋ, as well as for other Indigenous Peoples, this paper mostly incorporates Ojibwe perspectives, including the use of the Ojibwemowin word for wild rice, Manoomin⁶, with capitalization honoring its personhood¹⁵.

Two centuries ago, Manoomin was abundant across the shallow surface waters of the upper Laurentian Great Lakes region and portions of eastern North America^16,17, but today it has disappeared almost entirely from Michigan^18,19 and from one third of watersheds in Minnesota and Wisconsin where it was found about a century ago²⁰ (Fig. 1). Over the last few decades, tribes and inter-tribal organizations have reported continued declines of Manoomin^21,22; however, detecting recent multi-year trajectories of change and associated impacts on tribal harvesters has been elusive against a backdrop of Manoomin’s natural multi-year cycles²³. Further, although tribal-led studies have identified both land-use and climate change to be among the general causes for this decline^8–10,18, several tribal governments and organizations still seek quantitative assessments of specific relationships between climate conditions and Manoomin abundance.

This study quantifies declines in Manoomin density and harvest over the last three decades, ties these losses to climate disruptions during Manoomin’s vulnerable life stages, and shows threats to Indigenous lifeways. To accomplish this, we employ environmental science within both the mainstream (‘Western’) knowledge system and the Ojibwe gikendaasowin (Ojibwe knowledge system). Environmental science within Ojibwe gikendaasowin, also referred to here and more generally across different Indigenous knowledge systems as Traditional Ecological Knowledge (TEK), is science embedded in an Indigenous worldview developed through generations of Indigenous Peoples living in close relationship to their homelands^15,24,25. When invoked together, mainstream and Ojibwe gikendaasowin deepen our understanding of the Manoomin socio-ecological system in ways unattainable from each one alone^26–28. Given the dominance of mainstream environmental science within the academy, we strove for equitable collaboration by learning from Indigenous methodologies²⁹ and utilizing TEK as the ‘intellectual scaffold’ for our approach^30,31. Specifically, the fundamental understanding of interconnectedness within TEK directed our tribal-university partnership formation^32,33 and our methodological assumption people–including our knowledge production and colonial power dynamics–are inextricably linked with the environment^15,34.

With this methodological grounding, we leveraged local observations, understandings, values, and data from within TEK to generate relevant research questions and hypotheses, inform the appropriate choice of statistical methods, and draw meaningful interpretations of quantitative model outputs (Methods). Tribal priorities and TEK informed data collection: two inter-tribal organizations, the 1854 Treaty Authority (1854TA)²² and the Great Lakes Indian Fish and Wildlife Commission (GLIFWC)²¹ (Figs. 1 and 2), collected, respectively, 20 and 34 years of Manoomin density across a total of 49 historical Manoomin waterbodies in Wisconsin and Minnesota. GLIFWC additionally collected 27 years of harvest data across Wisconsin. The density data were assessed for periodicity using spectral analysis³⁵ and for temporal trends and relationships with climate using generalized additive models (GAMs)³⁶. Specifically, we hypothesized winter, spring, and early summer conditions would impact Manoomin growth, given Manoomin’s observed sensitivity during its early life stages⁶. The results show climate change—increasing early summer precipitation, higher mean winter temperatures, decreased snowfall, and shorter ice-duration across this region—negatively impacts Manoomin density, contributing to declines in harvest and thus threatening Indigenous lifeways.

To understand Manoomin’s varying abundance through time, both fluctuations and persistent trends in Manoomin density were characterized. The multi-taper method³⁵ was used to analyze potential cycles of Manoomin density for each waterbody (n = 49) given this method’s ability to handle short time series. Out of 49 waterbodies, 67% had significant multi-year cycles of density ranging from 2–7 years (median cycle = 3 years), and 41% had multi-decadal cycles (median cycle = 20 years) (Supp. Table 1).

GAMs were used to test for significant regional trends in density through time. GAMs were chosen because they can represent non-linear, place-specific, and global relationships^36,37 (Methods). The GAMs captured each waterbody’s unique temporal patterns and overall regional trends, explaining 70% and 66% of the variance in density for the 1854TA and GLIFWC datasets, respectively; 34% for the GLIFWC Wisconsin tribal harvest; and 23% for the GLIFWC Wisconsin non-tribal harvest (Fig. 3). The regional trend of Manoomin density for both datasets was a statistically significant trend over the last three decades of -6.8% (95% CI [-10.6, -2.8]). and − 5.6% (95% CI [-12.0, 1.3]) density change per year on average for the GLIFWC and 1854 Treaty Authority datasets, respectively. Most of this decline was concentrated in the mid-1980s through mid-2000s. In Wisconsin, off-reservation tribal harvest exhibited a statistically significant change of -2.0% (95% CI [-6.4, 2.7]) harvest per trip per year. Non-tribal harvest has not declined over the 1992 to 2019 analysis period but dipped during 2000–2010.

The hypotheses about regional Manoomin-climate relationships tested were derived from TEK and constrained by data availability. Specifically, local observations within TEK⁶ show connections between longer (multi)decadal periodicities of Manoomin abundance and oscillations of spring and early summer precipitation, and between cold, snowy winters and high Manoomin abundance (Methods). Four hypotheses emerged: higher precipitation during the submerged and floating leaf stages (spring and early summer) decreases density, more days with a minimum temperature above 4˚C (growing degree days) increase density, higher mean winter temperature decreases density, and greater snow accumulation increases density (Table 1).

To test these hypotheses, we assessed Manoomin density through time and in relation to climate datasets³⁸ using GAMs (Methods, Fig. 4). For both the GLIFWC and 1854TA datasets, higher precipitation in spring and early summer, during Manoomin’s floating leaf stage, were associated with lower density (GLIFWC: p < 0.001; 1854TA: p < 0.001). Precipitation during spring, Manoomin’s submerged stage, also had a significant correlation with Manoomin density, but this was a positive correlation for the GLIFWC dataset (p < 0.001) and negative for the 1854TA dataset (p = 0.007). The effects of growing degree days were also inconsistent across datasets: they had a significant negative correlation with Manoomin density in the GLIFWC dataset (p = 0.04) but no significant effect for the 1854TA dataset (p = 0.09). Mean winter temperature had a significant negative correlation with Manoomin density for the 1854TA dataset (p = 0.04) but the correlation for the GLIFWC dataset was just above the significance threshold (p = 0.06). The regional and water-body specific trends through time were significant in both models (GLIFWC p < 0.001; 1854TA p < 0.001). The effects of snow were consistently positive but not significant for both datasets. The models explained 76% and 74% of the variance in the GLIFWC and 1854TA data, respectively.

We further explored in-lake impacts of floating-leaf stage (early summer) precipitation and dormant stage (winter) conditions on Manoomin density using water-level data²² for the waterbodies monitored by 1854TA, and ice-duration data³⁹ for lakes nearby the waterbodies monitored by GLIFWC, based on data-type availability (Methods, Fig. 4). The results show shallower water levels during the floating-leaf stage (p < 0.001) and longer ice-duration (p < 0.001) were both positively correlated with Manoomin density. Water levels for the other growing stages did not have statistically significant effects. Both models explained 76% of variance in the data.

Changes in climate over time were also assessed with GAMs. All weather variable trends were significant (p < 0.05) since 1985 (Fig. 5). We extended the analysis for four records that went back to1900, and all weather variable trends were significant (p < 0.05) except for growing degree days (Fig. 5). Over both time periods, snowfall and precipitation have increased in dormant and submerged stages respectively. Precipitation during the floating-leaf stage has not increased on average, but the magnitude of peaks within the regular multi-year oscillations have increased over the last few decades.

Lake ice-duration and water levels temporal trends were also evaluated with GAMs. Generally, ice-duration has declined, especially in the late 1990s, and water levels in submerged and floating-leaf stages have increased since 1998 (Fig. 6). All exhibit substantial multi-year oscillations as well.

Informed by TEK (Methods), we hypothesized Manoomin harvested by tribal members relates positively to Manoomin density. Furthermore, we hypothesized the preceding years' harvest would not negatively impact Manoomin density, as suggested by TEK^6,13 and past studies^40,41, and could possibly even improve future harvest.

To assess these hypotheses, two more GAMs were developed: one investigating the relationship between the amount harvested by tribal members and Manoomin density, and another investigating the relationship between current-year and previous harvests (Methods). The first GAM showed a positive correlation between Manoomin density and tribal harvest per trip up to 250 stalks/m², after which harvest declined (Fig. 7). It also revealed a significant decline in harvest per trip through time. In the second GAM, the average harvest per trip of the previous two years had a significant positive correlation with harvest in the current year. In total, the GAMs explained 19% and 52% of total variance, respectively.

The combination of mainstream scientific and TEK monitoring methods provided a robust characterization of Manoomin abundance. The 1854TA approach privileges mainstream science through extensive ground surveys using unbiased, plot-level measurements that yield higher precision of fewer sites with greater time costs. Monitoring conducted by GLIFWC privileges TEK by employing multiple faster, lower-cost methods of aerial photos, harvester surveys, and simplified ground surveys that allow for long-term and widespread measures to cover a greater number of culturally important sites, even if with lower precision²⁶. Because of the complementary strengths of these methods, each dataset reveals different nuances regarding Manoomin abundance, and together they provide robust information about status and trends. Guided by TEK, both inter-tribal organizations focused on one plant and important ‘hot spots’ of harvest rather than randomized site selection across multiple species²⁶. Therefore, these data do not represent overall ecosystem health or total Manoomin abundance across the region, but instead, regional, off-reservation Manoomin available for Ojibwe harvest.

The short time series lengths (< 40 years, 1 datapoint/year) may weaken the robustness of the multi-year and (multi)decadal oscillations found in Manoomin abundance across the region with spectral analysis, but these variations are consistent with TEK^6,23 and experimental evidence^42,43 of natural cycles. Considering these natural fluctuations and site-level variation, GAMs revealed a statistically significant decline of ~ 5–7% decline per year in off-reservation Manoomin available for Ojibwe harvest across the region and ~ 3% per year decline in off-reservation harvest per trip for tribal members. The within-order-of-magnitude declines suggest they could be related. This harvesting trend may be impacted by demographically aging harvesters’ decreased productivity but increased skill, tribal members’ on-reservation harvest, and/or low sample size⁴⁴.

The significant negative correlation found between precipitation during the floating-leaf stage and Manoomin density for both datasets–despite their differences in geography and sampling methods–suggests a robust relationship. Further, the significant negative correlation between high water levels during Manoomin’s floating leaf stage and Manoomin density in the 1854TA dataset is consistent with our hypothesis, prior studies^45,45–49, TEK⁶, and historic Ojibwe trade records¹⁶. High precipitation generally raises water levels, which increases the energy required for Manoomin to reach the surface and photosynthesize⁵⁰ and can drown emerging plants^40,51. Also, high water levels decrease seed production⁴⁵ by delaying phenology^16,50,52 and increasing favorability of perennials over Manoomin⁵⁰. Together, these results suggest high precipitation during the floating-leaf stage has caused declines in Manoomin density by raising water levels.

The inconclusive correlation of both precipitation during the submerged-stage and water levels with Manoomin density across datasets could be because, during this life stage, plants are less susceptible to drowning or uprooting. Also, higher water levels in the spring submerged stage from spring rains and/or snowmelt may actually benefit Manoomin, as suggested by our GLIFWC model results (Fig. 4A), because they mix and flush sediments through rice beds, adding nutrients and uncovering seeds, both of which foster initial growth and stabilize long-term abundance^50,53. However, a correlation with snowfall was not seen for the 1854TA data, suggesting snowfall has an inconsistent, spatially complex, or weak relationship with Manoomin density. Both precipitation and water levels during the emergent stage were consistently unrelated to Manoomin density, likely because during the emergent stage, plants stand above the water’s surface and are physically robust. The lack of precipitation and water-level effects for this phenological stage further supports the TEK understanding that earlier life stages are more sensitive to precipitation and water level.

The negative correlation found with winter temperature for the 1854TA density data is consistent with observations that Manoomin seeds require seven months at 2.5-5˚C to fully break dormancy and germinate⁵⁴ and can withstand temperatures down to and possibly below − 10˚C^54,55. These results together suggest warmer winters likely cause fewer Manoomin seeds to germinate the following year. In addition, colder temperatures may benefit Manoomin relative to perennial plants less tolerant of extreme cold⁵⁶. Although the GLIFWC density data lacked a winter temperature correlation, it showed a significant positive correlation with nearby ice-duration. Both colder temperatures and longer ice-duration generally correspond to thicker lake ice⁵⁷. Thicker lake ice increases the possibility of ice-scouring, which mixes sediment, bringing buried seeds closer to the lake-bed surface where they can more readily germinate^16,58. While our results point to multiple processes through which Manoomin may be vulnerable to milder winters, correlations between growing degree days and Manoomin density were not consistent, suggesting weak or complex relationships.

The quantification of climate, water-level, and ice-duration relationships with Manoomin are complicated by these variables’ changes through time (Fig. 5), which are not accounted for in our models. The relationships between Manoomin density and climate, water level, and ice-duration have the potential to be spurious correlations; however, the multi-dataset and multivariate lines of evidence support Manoomin’s positive relationship with lower early summer precipitation, lower early summer water levels, and colder winters with more snow and longer ice duration. It is also possible time is the variable spuriously correlated, or partially so, and Manoomin’s relationship with climate, water level, and ice-duration are therefore underestimated within this analysis. Additionally, many relationships likely influencing Manoomin density and harvest were excluded from this model, such as lake and watershed morphology, watershed and shoreline land-use, herbivory, diseases, permitting processes, reseeding, restoration, and changing phenology⁵⁹.

We did include harvest into our analysis and found the average harvest from the previous two years had a positive relationship with harvest of a given year, suggesting human harvest may be beneficial for Manoomin populations, supporting our hypothesis, TEK^6,13, and past studies^40,41. By removing some of the seeds, harvesting may help thin out beds, possibly reducing local humidity and therefore the occurrence of brown spot fungal disease⁶⁰, or reducing the amplitude of abundance fluctuations and thus the opportunity for perennial vegetation to establish in Manoomin habitat during low-abundance years. Manoomin may also benefit from the disturbance caused by harvesters pushing through the rice beds, including agitation of sediments to bring buried seeds to the lakebed surface, and knocking seeds into the water that might have otherwise been eaten by birds. Additionally, harvesters may focus their stewardship efforts on the waterbodies from which they harvest the most, supporting greater future abundance. Because of this potential relationship, both the aging and decreasing number of harvesters over the last several decades²⁰ may additionally contribute to the decline of Manoomin due to loss of harvesting knowledge and practice.

Trends we detected in climate variables are consistent with those previously reported for the region beyond natural oscillations, including increasing winter temperature, increasing spring and summer precipitation totals and variation⁶¹, and decreasing ice-duration^62,63. Given the relationships we found between these climate variables and Manoomin density, we conclude climate change, specifically increases in early summer precipitation, warmer winter temperatures, and decreases in lake ice-duration, has already been impairing Manoomin density. Longer climate oscillations, as seen in decadal water-level cycles observed in the region^64,65, may also contribute to changes in Manoomin density, as suggested by TEK⁶⁶, in addition to the overlying signal of climate change.

Looking to the future, this region is predicted to receive increases in spring and early summer precipitation due to anthropogenic climate change beyond natural oscillations^11,67. Winter precipitation is expected to increase, but with greater partitioning to rain^11,68, and lower and earlier snowmelt⁶⁹. Major Manoomin regions—central Minnesota and the shores of Lake Superior—will experience the sharpest declines in snowfall across the region, reaching about 50% reduction by the middle of the 21st century⁶⁸. The warming trend in mean winter temperatures over the last half of the twentieth century⁶¹ is predicted to continue and increase ^11,67,68. Ice-duration is projected to further decline by approximately 20 days by 2070⁶³ and 40 days by the end of the century^63,70. Given the correlations found between these climate factors and Manoomin density, we expect these projected changes will further threaten Manoomin.

The decline of Manoomin, in part driven by climate change, has and will continue to negatively affect tribal communities. Results show Wisconsin tribal harvest positively corresponds with Manoomin density, suggesting declines in Manoomin likely contribute to the lower Wisconsin tribal off-reservation harvest rates since 1992. Very high Manoomin density (> 250 stalks/m²) was associated with fewer kilograms harvested per trip, likely because of difficulties navigating very dense beds during harvest⁷¹. Climate change may also affect harvest in ways unrelated to stalk density: extreme low water levels in August can prevent canoe access to beds for harvesting¹⁶; storms can prematurely knock rice off of stalks ^19,42; hot, low-wind conditions during the flowering stage may limit pollination for seed production^16,40 and increase prevalence of brown spot fungal disease⁶⁰; excessively hot conditions during harvest can prevent people from participating; and all of this variability can make planning for harvest difficult.

The decline in Manoomin access over the last decades has disrupted Ojibwe lifeways, family, and health^72,73. Manoomin cannot simply be replaced by another food because it is an “integral glue” connecting political, economic, social, spiritual, intellectual, and physical dimensions of Anishinaabeg society⁷⁴. This loss therefore fails to uphold the inherent rights to self-determination and sovereignty⁴ of these Ojibwe Nations, as recognized and explicitly promised by treaties with the United States^6,75,76 and more broadly guaranteed by the UN Declaration of the Rights of Indigenous People³. Ojibwe Nations continually address this loss through stewardship, research, partnership, and legal action^32,74,77. Indigenous Peoples from across the world are also experiencing loss of traditional foods, and their connected culture, knowledge systems, and sovereignty, all driven in part by climate change^78–80. These are the very people who lead in climate stewardship through mitigation, sequestration, and storage of carbon^81–83 over one quarter of the world’s land surface⁸⁴. The loss of Manoomin thus contributes to the erosion of humanity’s collective biocultural diversity essential to producing the transformative knowledge for climate solutions^{1,78,82,84–88}.

ACKNOWLEDGEMENTS

I would like to acknowledge the Kawe Gidaa-naanaagadawendaamin Manoomin Collaborative for guiding and supporting this work. Specifically, I would like to acknowledge Peter David for his decades-long efforts in monitoring Manoomin at GLIFWC. His data and knowledge directly enabled and improved this research.

Fernández-Llamazares, Á. et al. Scientists’ Warning to Humanity on Threats to Indigenous and Local Knowledge Systems. J. Ethnobiol. 41, (2021).
United Nations. United Nations Declaration on the Rights of Indigenous Peoples. (2008) doi:10.1111/j.1467-8322.2008.00569.x.
NCAI. Tribal Nations and the United States: An Introduction. (2020).
David-Chavez, D. M. & Gavin, M. C. A global assessment of Indigenous community engagement in climate research. Environ. Res. Lett. 13, (2018).
David, P., David, L., Stark, H. K., Fahrlander, S. N.-A. & Schlender, J. M. Manoomin, Version 1.0. 158 (2019).
GLIFWC Climate Change Team. Aanji-bimaadiziimagak o ’ ow aki: Climate Change Vulnerability Assessment Version 2. (2023).
Byun, K. & Hamlet, A. F. Projected changes in future climate over the Midwest and Great Lakes region using downscaled CMIP5 ensembles: PROJECTED CLIMATE CHANGES OVER THE MIDWEST AND GREAT LAKES REGION. Int. J. Climatol. 38, e531–e553 (2018).
Kimmerer, R. W. Braiding Sweetgrass. Milkweed Ed. 248 (2013).
Vennumn, T. Wild Rice and the Ojibway People. (Minnesota Historical Society Press, St. Paul, MN, 1988).
Smart, A. The Ecological Requirements of Wild Rice (Zizania Spp.) in Michigan. (Central Michigan University, 2013).
Drewes, A. D. & Silbernagel, J. Uncovering the spatial dynamics of wild rice lakes, harvesters and management across Great Lakes landscapes for shared regional conservation. Ecol. Model. 229, 97–107 (2011).
David, P. F. Manoomin (Wild Rice) Abundance and Harvest in Northern Wisconsin in 2014. (2020).
Vogt, D. J. Wild Rice Monitoring and Abundance in the 1854 Ceded Territory (1998-2022). 32 http://1854treatyauthority.org/cms/files/REP WR Monitoring 2012.pdf (2023).
Kimmerer, R. W. The Fortress, the River and the Garden. in Contemporary Studies in Environmental and Indigenous Pedagogies (eds. Kulnieks, A., Longboat, D. R. & Young, K.) 49–76 (Sense Publishers, Rotterdam, 2013). doi:10.1007/978-94-6209-293-8_4.
Moller, H., Berkes, F., Lyver, P. O. & Kislalioglu, M. Combining Science and Traditional Ecological Knowledge: Monitoring Populations for Co-Management. Ecol. Soc. 9, (2004).
Matson, L. et al. Transforming research and relationships through collaborative tribal-university partnerships on manoomin (wild rice). Environ. Sci. Policy (2021).
Mann, M. & Lees, J. Robust estimation of background noise and signal detection in climatic time series. Clim. Change 33, 409–445 (1996).
Wood, S. Generalized Additive Models: An Introduction with R. (CRC Press, 2017).
NOAA. National Weather Service Cooperative Network. (2023).
NSIDC. National Snow and Ice Data Center. (2023).
Moyle, J. B. Wild Rice in Minnesota. J. Wildl. Manag. 8, 177 (1944).
Durkee Walker, R. E., Pastor, J. & Dewey, B. W. Litter quantity and nitrogen immobilization cause oscillations in productivity of wild rice (Zizania palustris L.) in northern Minnesota. Ecosystems 13, 485–498 (2010).
David, P. Personal Communication. (2023).
Archibold, O. W. & Weichel, B. J. Variation in wild rice (Zizania palustris) stands across northern Saskatchewan. Can. J. Bot. 64, 1204–1211 (1986).
Meeker, J. E. The Ecology of ‘wild’ wild-rice (Zizania palustris var. palustris) in the Kakagon Sloughs, a riverine wetland on Lake Superior. in Wild Rice Research and Management Conference (eds. Williamson, L. S., Dlutkowski, L. A. & Mccammon Soltis, A. P.) 68–83 (Great Lakes Indian Fish and Wildlife Commission, 1999).
Kovach, D. A. & Bradford, K. J. Temperature Dependence of Viability and Dormancy of Zizania palustris var. interior Seeds Stored at High Moisture Contents. Ann. Bot. 69, 297–301 (1992).
Adamec, L. & Kučerová, A. Overwintering temperatures affect freezing temperatures of turions of aquatic plants. Flora - Morphol. Distrib. Funct. Ecol. Plants 208, 497–501 (2013).
Karetnikov, S., Leppäranta, M. & Montonen, A. A time series of over 100 years of ice seasons on Lake Ladoga. J. Gt. Lakes Res. 43, 979–988 (2017).
Counts, R. L. Phenotypic plasticity and genetic variability in annual Zizania spp. along a latitudinal gradient. Can. J. Bot. 71, 145–154 (1993).
R. Nyvall. Comparison of fungal brown spot severity to incidence of seedborne Bipolaris oryzae and B. sorokiniana and infected floral sites on cultivated wild rice. Plant Dis. 79, 249–250 (1995).
Pathak, P., Kalra, A. & Ahmad, S. Temperature and precipitation changes in the Midwestern United States: implications for water management. Int. J. Water Resour. Dev. 33, 1003–1019 (2017).
Newton, A. M. W. & Mullan, D. J. Climate change and Northern Hemisphere lake and river ice phenology from 1931–2005. The Cryosphere 15, 2211–2234 (2021).
Hewitt, B. et al. Historical Trends, Drivers, and Future Projections of Ice Phenology in Small North Temperate Lakes in the Laurentian Great Lakes Region. Water 10, 70 (2018).
Watras, C., Latzka, A. W., Meyer, M. W. & Zhang, Z. Near-Decadal Oscillation of Water Levels and Mercury Bioaccumulation in the Laurentian Great Lakes Region. (2020) doi:10.1021/acs.estlett.9b00772.
Xue, P. et al. Climate projections over the Great Lakes Region: using two-way coupling of a regional climate model with a 3-D lake model. Geosci. Model Dev. 15, 4425–4446 (2022).
Liess, S. et al. High-Resolution Climate Projections Over Minnesota for the 21st Century. Earth Space Sci. 9, 1–16 (2022).
Demaria, E. M. C., Roundy, J. K., Wi, S. & Palmer, R. N. The effects of climate change on seasonal snowpack and the hydrology of the Northeastern and Upper Midwest United States. J. Clim. 29, 6527–6541 (2016).
Schuldt, N. et al. Expanding the Narrative of Tribal Health: The Effects of Wild Rice Water Quality Rule Changes on Tribal Health Fond du Lac Band of Lake Superior Chippewa Health Impact Assessment. (2018).
Whyte, K. P. Renewing Relatives: Indigenous Food Sovereignty and Settler Colonialism. (2016).
Hernandez, J. Fresh Banana Leaves: Healing Indigenous Landscapes through Indigenous Science. (North Atlantic Books, 2022).
Waller, D. M. & Reo, N. J. First stewards: Ecological outcomes of forest and wildlife stewardship by indigenous peoples of Wisconsin, USA. Ecol. Soc. 23, (2018).
Garnett, S. T. et al. A spatial overview of the global importance of Indigenous lands for conservation. Nat. Sustain. 1, 369–374 (2018).
GLIFWC. WI Manoomin Inventory, V2.2.1. (2023).
MNDNR. Wild Rice Lakes Identified by DNR Wildlife. (2020).
MN DNR. DNR Hydrography Dataset. (2023).
WI DNR. 24k Hydro Full File Geodatabase. (2022).
USFS. Tribal Lands Ceded to the United States (Feature Layer). (2018).
Khoury, C. K. et al. Crop wild relatives of the United States require urgent conservation action. Proc. Natl. Acad. Sci. 117, 33351–33357 (2020).
Native Land Digital. Native Land Digital. (2023).
Kjerland, T. Wild Rice Monitoring Handbook. The University of Minnesota Sea Grant Program. (2015).
Warne, C. P. K., McCann, K. S., Rooney, N., Cazelles, K. & Guzzo, M. M. Geography and Morphology Affect the Ice Duration Dynamics of Northern Hemisphere Lakes Worldwide. Geophys. Res. Lett. 47, e2020GL087953 (2020).
Rahim, K. multitaper. (2020).
Ross, N. GAMs in R. (2019).
Vittinghoff, E., Glidden, D. V., Shiboski, S. C. & McCulloch, C. E. Regression Methods in Biostatistics: Linear, Logistic, Survival, and Repeated Measures Models. (Springer US, Boston, MA, 2012). doi:10.1007/978-1-4614-1353-0.

Collaborative Process

This research was conducted within the Kawe Gidaa-naanaagadawendaamin Manoomin Research Collaborative, a tribal-university partnership with formalized agreements between four Ojibwe Nations (Fond du Lac Band of Lake Superior Chippewa, the St Croix Indians of Wisconsin, the Mille Lacs Band of Ojibwe, and the Lac du Flambeau Band of Lake Superior Chippewa), two inter-tribal organizations (the Great Lakes Indian Fish and Wildlife Commission and the 1854 Treaty Authority), and a group of interdisciplinary researchers at the University of Minnesota-Twin Cities³². From the first three years of collaboration (2018–2020), assessing the impact of climate change on Manoomin abundance was determined to be a collective priority across the tribal partners, especially given the high vulnerability of Manoomin to climate change identified by tribal climate adaptation plans^8–10.

The specific hypotheses for climate change impacts on regional Manoomin abundance were developed from TEK shared through tribal documentation^6,8, monthly project meetings, (semi-)annual project conferences, knowledge-exchange workshops, and our time together paddling through rice beds, harvesting, visiting, ceremony, and feasting. In addition, we hosted 18 virtual meetings to share and discuss water level, water chemistry, and vegetation data from specific Manoomin waters across the region over the winters of 2021 and 2022. These meetings included 19 tribal resource managers from 6 (inter-)tribal organizations across the region, 2 state resource managers, and 18 university researchers, students, and staff. In the summer of 2022, 35 knowledge-holders representing tribal and state agencies, tribal and non-tribal rice harvesters, academic institutions, and non-governmental organizations convened for a 3-day Tribal Adaptation Menu⁹⁶ Workshop on caretaking strategies for Manoomin in the face of climate change. Participants voted on the most pressing climate change impacts on Manoomin in their waterbody of concern (Supp. Figure 1). All of his knowledge along with data availability constraints were iteratively refined into testable hypotheses for this study. Additionally, this knowledge informed the choice of methods: statistical tools to assess periodicity and non-linear relationships, and account for unique relationships within each waterbody.

Following the initial modeling process, results were discussed with tribal partners during formal and informal visits, updates at monthly meetings, and via email. Preliminary results were then shared, and the modeling was refined through discussions with the larger tribal natural resource and Indigenous research community in addition to the mainstream scientific communities at local, regional, and national conferences. Specifically, variable calculations were adjusted based on additional local observations within TEK, more models were created to look at the relationships in new ways, and insights and limitations of the results were revealed. This iterative process began in the fall of 2021 and continued through the fall of 2023.

Datasets

Manoomin Density

Data were obtained from the 1854 Treaty Authority (1854TA) and the Great Lakes Indian Fish and Wildlife Commission (GLIFWC). Each organization has been monitoring Manoomin abundance across their respective geographic regions for several decades. The 1854TA monitored Manoomin density and biomass on 9 lakes from 1998 to present by averaging Manoomin stalks and plant height in 20–25 0.5-m² plots in each lake^22,97. Their sampling method changed in 2006 when they transitioned from randomized plot locations within a rice bed to a uniform, gridded set of plots²². We assumed this transition from biased to unbiased sampling would influence the data, and therefore included a random-effects term to account for this change by allowing for different slopes and intercepts for each sampling method.

GLIFWC has been regularly monitoring Manoomin on 40 waters from 1985 to 2019 through a combination of aerial surveys and on-the-water density measurements²¹. Aerial photos were taken with aircraft-based, hand-held cameras at oblique angles. Boundaries of Manoomin stands in each photo were plotted by hand onto waterbody maps to estimate the total acreage of Manoomin⁴⁴. Given the hand-adjusted oblique angle and difficulty of delineating precise stand extent from an aerial photo, these acreage measurements should be considered estimates with significant error. Manoomin density measurements were made by counting stalks in 20 randomly selected 0.5-m² plots within a Manoomin bed on each lake. These values were then binned into five categories: (1) 0–20, (2) 20–40, (3) 40–60, (4) 60–80, and (5) 80 + average stems per plot. For this analysis, the median number of stalks per plot was used for each bin (bin 1 = 10 stems/plot, bin 2 = 30 stems/plot, bin 3 = 50 stems/plot, bin 4 = 70 stems/plot, and bin 5 = 90 stems/plot). These binned data reduce variability and do not accurately capture densities above 80 stems per plot. Each stalk density was then multiplied by the Manoomin acreage and divided by the entire waterbody area to calculate an average density.

Although the overall sampling method used by GLIFWC contains more error at the site level than the sampling method used by the 1854TA, the lower-effort and simplicity of the method enabled GLIFWC to maintain a consistent survey of Manoomin abundance of 40 lakes for 34 years, providing an unprecedented dataset for which trends can be analyzed at a regional and multi-decadal scale with fair confidence. Equations relating Manoomin density and plant height to biomass have been developed, but were not used because they would introduce additional assumptions and exclude GLIFWC data that lacked plant height. In total, we had 1507 unique data points across 49 waters across the two datasets to include in our analysis.

Manoomin Harvest

GLIFWC surveyed state (non-tribal) and tribal harvesters annually in Wisconsin from 1992 to 2019. Harvesters were identified by registrations for rice harvesting permits, required for all non-tribal harvesters and tribal harvesters ages 17–64 harvesting off-reservation. All harvesters who had reported harvest the previous year and a randomly selected half of those who were registered but did not report a harvest the previous year were contacted via phone or mail to determine the number of trips, kilograms harvested, and harvesting locations that year off-reservation. The decline in harvest totals may be due to a decline in survey response rate reported by GLIFWC staff. Due to this phenomenon, we used kilograms per trip harvested as the metric of analysis to quantify the amount of harvest per ‘unit effort’²⁶ given that declining response rate made it impossible to accurately determine total harvest. In total, we have 2514 unique harvesting data points included in our analysis.

Climate

Daily precipitation, snow accumulation, and maximum and minimum daily air temperature data for 21 different weather stations in the National Weather Service Cooperative Network³⁸ were obtained from the Midwest Regional Climate Center data portal. The stations were chosen based on closest proximity to each of the Manoomin waters and their near-continuous records (> 80%) over the study period. These climate variables were aggregated into total precipitation within three life stages of Manoomin (submerged [Julian day 92–150], floating [Julian day 151–199], flowering/emergent [Julian day 200–229]), number of days in the growing season (Julian day 92–229) with minimum temperature above 4˚C (growing degree days), average preceding winter temperature, and total snow accumulation in the preceding winter (Table 1). Manoomin phenology of life stages has likely been changing with climate, which was not accounted for in this analysis. However, all data were aggregated over the approximate life stage time, and so we assume that small changes in start and end dates of each life stage would have only minor impacts on aggregated totals or averages.

Ice Cover

Because ice-on and ice-off dates were not recorded for the Manoomin waters themselves, ice-on and ice-off data for the closest lake with data extending back to 1985 (with a few missing years) were obtained from the National Snow and Ice Data Center (Supp. Table 1). Ice-duration was calculated by counting the days between ice-on and ice-off records. Ice-duration was used as a predictor, rather than ice-on or ice-off dates themselves, because it integrates both freeze-up and break-up date changes into one metric. The exact ice-duration of Manoomin waterbodies was probably different from these nearby lakes, because morphological differences between lakes, namely depth, impact ice-on and ice-off timing. However, nearby lake ice-durations were considered justifiable proxies for Manoomin waterbody ice-duration, because with similar climatic and watershed conditions based on proximity, they can be expected to experience similar relative changes through time. Lake ice-on and ice-off was used from Crystal Lake, Allequash Lake, Shell Lake, Lake Escanaba, and Trout Lake, and were limited only to the 1837 and 1842 ceded territories monitored by GLIFWC due to limited data in other regions (Supp. Table 1).

Water Levels

Bi-weekly water levels were recorded by 1854TA for their 10 Manoomin waters of study from 1998 to 2020²². Water level anomalies (deviations from time-average values for each growing stage for each lake over the entire study period) were used to capture relative water level, because water level gauges were not always located in the Manoomin beds.

Spectral Analysis

To assess the relatively short time series (20–34 data points), we performed the multi-taper method (MTM) on each time series individually to determine significant periodicities³⁵. MTM efficiently uses all data and minimizes spectral leakage, making it suitable for the spectral analysis of short datasets. Significance was assessed using an F-test, with frequency peaks over 95%, and not exceeding the time series length, considered significant. Frequencies were rounded to the nearest whole number of years for values less than 10 years and rounded to the nearest 10 for values greater than 10 because partial-year periodicities are not likely to occur ecologically, and there is significant error in the estimation of (multi)decadal periodicities given the short dataset. Calculations were performed using the multitape⁹⁹ package in R¹⁰⁰.

Generalized Additive Models

Generalized additive models (GAMs) were chosen to model Manoomin abundance and harvest trends through time, along with Manoomin-climate relationships, because of the inherent variability in abundance and harvest through time and the complex (nonlinear) and interconnected eco-hydrologic relationships identified through TEK. Like generalized linear models, GAMs can assess multiple covariate relationships simultaneously; however, relationships in GAMs are not constrained to be linear. GAMs use smaller basis functions, in this case cubic splines, added together to make an overall smooth function that can be highly non-linear (i.e., “wiggly”) to relate a variable of interest to multiple predictor variables. The extent of “wiggliness” is determined by the number of basis functions, a parameter set within the model, and a built-in penalty to avoid overfitting. With these characteristics, GAMs provide a flexible framework for modeling nonlinear relationships of Manoomin abundance and harvest with time and climate. Multiple GAMs were constructed to evaluate different hypotheses (see Table 1) using the mgcv¹⁰¹ package in R¹⁰⁰. All GAMs were assessed for appropriate number of basis functions (smoothness) and data distribution, and lack of significant autocorrelation and concurvity (a nonlinear analogue of collinearity). Recommended thresholds were only exceeded by concurvity values for the site-specific random effects and total snow (recommended threshold = 0.8). These predictors were retained in models given our ecological hypotheses and to reflect site-specific variability and the nested sampling design (nonindependence of data arising from multiple plots and repeat sampling of each waterbody).

The first set of GAMs included one global smooth curve with time to assess the overall density trend across the region, and a random-effects term with time to evaluate site-specific trends (Supp. Table 1). For the density data, the random effects were modeled using different smooth curves of response variables through time for each waterbody because this allows parameterization of the potential individual “wiggliness” for each site. The number of basis functions was set relatively high (k = 10) for both the overall trend and site-specific trends through time to account for the possibility of periodicity. Higher values were tested but did not improve fit. The harvesting dataset included data from 275 waterbodies but only a few data points for each, so there was not enough data per waterbody to model random effects for each with a separate smooth curve through time. Instead, independent y-intercepts for the predicted variable (harvest rates) and basis function coefficients were set for each waterbody. For this model of harvest data, the overall basis function number was set even higher at 20 to account for the potential of more periodicity in this term because site-specific periodicity within the data was not accounted for with a random-effects term. Given the difference in sampling methods, the 1854TA density dataset (n = 202) and the GLIFWC density dataset (n = 1305) were modeled separately, as well as the GLIFWC harvest dataset for tribal members (n = 1234) and non-tribal members (n = 1930) to distinguish the two populations. The average percent change per year was calculated from the derivatives of the smooth curves¹⁰³.

The second set of GAMs augmented the previous GLIFWC (n = 1049) and 1854TA (n = 188) models to include smooth curves for precipitation total in each of the submerged, floating leaf, and emergent life stages; snowfall and mean air temperature from the preceding fall and winter (October - March); and growing degree days (days above 4˚C preceding fall harvest). The third set of GAMs augmented these further to explore intermediary variables that relate climate and Manoomin density: water level data were added to the previous 1854TA GAM (n = 204), and nearby ice-duration datasets were added to the previous GLIFWC GAM (n = 1025) based on data availability. Water levels were averaged for each life stage. Precipitation and mean winter temperature were then removed from the 1854TA and GLIFWC GAMs, respectively, because these variables were too similar (exceeding concurvity threshold of 0.8) to the water level and ice-duration data.

The fourth set of GAMs evaluated relationships with kilograms of Manoomin harvested per trip. One GAM assessed the relationships between GLIFWC survey data of Manoomin kilograms per tribal harvester and Manoomin density in the same year, regional trend through time, and random effects for each waterbody (n = 273). The second GAM assessed the kilograms per trip through time at sites for both tribal and non-tribal harvesters with the previous two years of harvest, and the random effects for each waterbody, including all waterbodies with harvesting data, and not just the ones surveyed for Manoomin density (n = 1282). Lastly, in a fifth set, the climate variable trends were modeled with GAMs for each of the weather station datasets since 1985 (data points = 444). Four weather stations had data extending back to the early 1900s (data points = 664), so an additional set of GAMs was created to assess longer-term trends in these datasets (Fig. 4). Each GAM assessed regional trends through time (one single smooth curve for all sites) and site-specific trends (different smooth curves for each waterbody).

Table 1 is available in the Supplementary Files section.

There is NO Competing Interest.

Download PDF

Version 1

posted

You are reading this latest preprint version

Climate Change Contributes to Loss of Wild Rice (Manoomin/Psiŋ), Threatening Indigenous Lifeways

Status:

Version 1

Abstract

Figures

INTRODUCTION

Manoomin Trends over the Last Three Decades

Climate Impacts on Manoomin Density

Relevant Climate Changes for Manoomin

Relationships with Indigenous Harvest

DISCUSSION

Declarations

References

METHODS

Collaborative Process

Datasets

Manoomin Density

Manoomin Harvest

Climate

Ice Cover

Water Levels

Spectral Analysis

Generalized Additive Models

Table 1

Additional Declarations

Supplementary Files

Status:

Version 1