Catastrophic flooding effects on a Wisconsin wet prairie remnant: A shift in the disturbance regime?

Abstract

Climate change is likely to imperil native biodiversity through the increased frequency of extreme events. Here we address the short-term effects of an extreme flooding event on an unplowed prairie reserve, the Faville Prairie Wisconsin State Natural Area. This 25-ha property is a remnant of the formerly extensive Crawfish Prairie that lay on the east bank of the Crawfish River, Jefferson County, Wisconsin USA. The Faville remnant has historically been subject to late winter to spring flooding in its lower portions. In June of 2008, however, an extreme rainfall event caused flooding unprecedented in the 87-year history of streamflow, inundating the entire site. Data were available from 180 permanently marked plots sampled in 1978-79. We assessed the change by resampling these plots in 2010-2015. At the m2 scale, we found significant losses of species richness, a result of most species having fewer occurrences than in the earlier data. There was near extinction of several important prairie species and a relative increase in wetland tolerant species. Lower elevation plots, subject to the encroachment of woody plants and the invasion of Phalaris arundinacea for decades prior to the flood, had the lowest levels of species richness. However, some prairie species survived the flooding with little change, and recent anecdotal observations show that others are rebuilding their populations. Thus, if extreme floods are infrequent, the prairie should be able to recover to its former state. If, however, the hydrological regime shifts toward more frequent, growing-season floods, we predict further decline in those plant species that were the object of the preservation of this remnant. It is critical that fire management continue along with monitoring to track species' recovery or replacement, so that corrective measures can be identified and tested to sustain the native prairie species diversity.

Fig 1. Aerial view of the Faville Prairie showing its location next to the Crawfish River.

The black line indicates the borders. The arrow in the upper left points to North. The squares show the positions of the 180 permanently marked points established by Max Partch in 1947–48 which were the center of 4 m² plots. These points were resampled in 1978–79, partially in 2005 [14], and again for the entire set in this study. The parcel bordering the Faville Prairie on the south is undergoing restoration by the Madison Audubon Society.

Fig 2. Daily maximum average discharge (m³/sec) for each day of the year as measured at the USGS Milford Gauging Station on the lower Crawfish River (43°06’00" N, 88°50’58" W).

Data plotted here are drawn from 1932 to 2020. The 2008 flooding corresponds to the sharp peak at days 168–170 (17–19 June). A secondary high peak, at about day 96 (5 April) records high spring flows in 2019. Note that the usual time of maximum discharge is between about day 67 and day 123 (2 March to 2 May). The day numbers of the maxima are approximate after day 59 due to the presence of leap years.

Fig 3. Species richness for each of the 180 sample points for the 1978 (X axis) and 2010 (Y axis) samples.

Though highly variable, the overall relation is positive (dashed line, Y = 0.08 + 0.73X, R² = 0.43, p < 0.0001, N = 180), indicating that a correlation between times was preserved despite an overall loss of richness. The slope of the regression reflects the greater absolute losses in the higher richness points.

Fig 4

Contour plots of point species richness for the 1978 survey (left panel) and 2010 survey (right panel) for the entire 180 plot array. The plots are arranged with north at the top. Note the strong expression of the ridge in the data for both surveys. The 1978–2010 comparison also shows the expansion of the low diversity portions of the site, especially in the easternmost portions of the site, reflecting the general loss of alpha-level richness over the entire site. Note that the eastern limit of the point array is about 200 m short of the summer position of the Crawfish River.

Fig 5. The mean wetland indicator value (MWI, see methods), the 2010 survey (Y) against the MWI for the 1978 survey (X), by point (N = 180).

Species were assigned an integer value from 5 (Obligate wetland species) to 1 (Obligate upland species). The dashed line is a regression (Y = 1.19 + 0.74X, R ² = 0.63, p <0.0001.). The solid line is the line of no difference (i.e., Y = X). Most of the points lie above this line, indicating that they had increased mean values in 2010, that is, greater representation of wetland tolerant species.

Fig 6. Comparison of the mean values of the “coefficient of conservatism” for each point in the two surveys (N = 180).

Higher values indicate more value for biodiversity conservation. The dashed line is the regression (Y = 0.44 + 0.80X, R² 0.53, p < 0.0001) and the solid line a reference line of no difference).

Fig 7. The occurrences of the 89 species that were present in both samples.

Lumped taxa (e.g., Carex spp.) are excluded. The dashed line is the linear regression on the untransformed data, and the solid line the line of no difference. Although the variance around the regression line is large, the overall the trend is positive and significant (Y = 8.0 + 0.45X, R² = 0.30, p < 0.0001). The colors indicate the Wetland Indicator Status; Red = Obligate wetland, Green = Facultative wetland, Cyan = Facultative, Dark Blue = Facultative upland, Black = Upland.

Fig 8. Maps of individual species showing their distribution within the 180 points in 78–79 (small red points) and in 10–15 (open circles).

Note the high site fidelity shown in panels D and E. The lines in the P. arundinacea graph (panel I) indicate the split used to define the regularly flood affected portion of the sampling array (see text).

Fig 9. Natural logarithmic transformation of mean annual discharge (m³/sec) of the Crawfish River as recorded at the USGS Milford Gauging Station (1931–2019).

The data and the regression line have little value for predicting discharge for any single year, but a strong trend (R² = 0.34, p < 0.0001, N = 88). These data do not reflect seasonality of flows, which is important for determining flooding effects on the vegetation (Fig 2). The data are too few to conclude that growing season floods are increasing in frequency.