Temporal scaling of episodic point estimates of weed seed predation to long-term predation rates

Authors: Davis, Adam; DAEDLOW, D - University Of Rostock; Schutte, Brian; WESTERMAN, P - University Of Rostock

Submitted to: Methods in Ecology and Evolution
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/6/2011
Publication Date: 12/5/2011
Citation: Davis, A.S., Daedlow, D., Schutte, B.J., Westerman, P. 2011. Temporal scaling of episodic point estimates of weed seed predation to long-term predation rates. Methods in Ecology and Evolution. 2:382-692.

Interpretive Summary: Many insects, birds and rodents consume weed seeds as a major component of their diet. Seed predation by granivores helps to hold down weed population growth, an important ecosystem service that can contribute to weed management in low-external-input agroecosystems. For convenience, measurements of seed predation are often made at very short time scales (less than 3 days). However, one of the primary uses of such data, building models of cropping system management on weed population dynamics, requires estimates of weed seed predation at seasonal or annual time scales. The relationship between short- and long-term point estimates of seed predation has not previously been characterized, hindering weed modeling efforts. In order to quantify this relationship, a field study was made within a maize crop in central Illinois, USA, in 2005 and 2006. From late July through mid-October, weekly short-term measurements were made of velvetleaf, giant ragweed, and giant foxtail seed removal rates over 48 hours by invertebrate and vertebrate granivores. Next to each short-term predation arena, long term predation rates were also measured by adding seeds weekly in long-term arenas, but not recovering the seeds until the entire study period had elapsed. We found a simple 1 to 1 relationship between observed long-term seed predation rates and the total number of seeds eaten over the study period divided by the total number of seeds added in the short term predation arenas. Moreover, we found that replication could be reduced by 90% without distorting the fundamental relationship between short and long-term estimates of seed predation. Future applications of this method should permit accurate estimates of long-term seed predation rates in agroecosystems with minimal sampling effort.

Technical Abstract: Weed seed predation is an important ecosystem service supporting weed management in low-external-input agroecosystems. For convenience, measurements of seed predation are often made at very short time scales (< 3 d). However, one of the primary uses of such data, the parameterization of models of cropping system management on weed population dynamics, requires estimates of weed seed predation at seasonal or annual time scales. The relationship between short- and long-term point estimates of seed predation has not previously been characterized, hindering weed modeling efforts. In order to quantify this relationship, a field study was made within a maize crop in central Illinois, USA, in 2005 and 2006. From late July through mid-October, weekly short-term measurements were made of Abutilon theophrasti, Ambrosia trifida, and Setaria faberi seed removal rates over 48 hr by invertebrate and vertebrate granivores. Next to each short-term predation arena, long term predation rates were also measured by adding seeds weekly in long-term arenas, but not recovering the seeds until the entire study period had elapsed. Short term predation rates were then used to predict long-term predation rates by two methods: 1) compounding short term predation rates over the study period, and 2) dividing the total number of seeds eaten over the study period by the total number of seeds added. Mixed-effects analysis of covariance indicated that method 2 provided a robust means of estimating long term predation rates from short term predation rates for all species and site-years (regression intercept = 0.1, regression slope = 1.01, R-sq = 0.74, p<0.001). Moreover, resampling analyses of 96 experimental units consisting of seed predation point estimates across all weekly sampling periods indicated that replication could be reduced by 90% without distorting the fundamental relationship between short and long-term estimates of seed predation. Future applications of this method should permit accurate estimates of long-term seed predation rates in agroecosystems with minimal sampling effort.