Legume species were interseeded separately into endophyte-infected tall fescue at the Mound Valley Unit of the Kansas State University Southeast Agricultural Research Center (37˚11.5'N, 95˚27.4'W). The plot area had been seeded to tall fescue >2 years prior to the test, and had been hayed without benefit of fertilization. The soil was a Parsons silt loam (fine, mixed thermic Mollic Albaqualf) with a pH of 6.3, 7 mg∙kg⁻¹ of available P (Mehlich 3), and 128 mg∙kg⁻¹ of exchangeable K in the upper 15 cm, to which was added 25 kg/ha of P and 28 kg/ha of K. Spring temperature and precipitation data for the location in 2010 and 2011 are listed in Table 1.

Species were planted with a no-till plot seeder (6 rows with 25-cm spacings, at ~5 mm depth) in 6-m strips, 12 m long in four replications, within a randomized strip split-plot design. Seeding dates for the experiment were April 20, 2010, and April 18, 2011, in adjacent areas. Seeding rates of the species and cultivars provided approximately 250,000 seeds∙ha⁻¹ and are listed in Table 2.

The seeded strips of legumes were randomly split into three 2-m subplots, each with six 25-cm rows, to determine the effects of pesticide protection on establishment. Grass suppression treatments (Table 3) were applied in 4-m strips randomly across each replication perpendicular to pesticide treatments, providing individual subplots of 1.5 by 4 m.

Legume seedling counts were taken randomly within a rectangle 33 cm long across the two center rows. In 2010, the rectangle was tossed about midway through the plot length; in 2011, two rectangles were counted at about one-third and two-third of the plot length, respectively. Legume seedling counts were made on June 17, 2010, and June 1, 2011, corresponding to 58 and 50 days after planting (DAP), respectively.

Data were analyzed using PROC MIXED in SAS [16] , with replication considered random and all other factors fixed. Fisher’s protected LSD was used for multiple comparisons (α = 0.05).

The same three legume species (Table 2) were mixed and interseeded as previously described into endophyte-

infected tall fescue that had been managed as pasture, including fertilization, for >5 years at the Parsons Unit of the Southeast Agricultural Research Center (37˚21.6'N, 95˚17.3'W) on Parsons silt loam. Spring temperature and precipitation data for the location in 2012 and 2013 are listed in Table 1. Plots (9.1 by 3.7 m) were seeded as previously described, in separate experiments on February 27, 2012, and March 15, 2013, with the cultivars and the rates listed (Table 2).

Suppression treatments were 1) grazing in lieu of clipping, 2) chemical suppression, and 3) an untreated check, randomized in four blocks. Heifers (>225 kg) of primarily Angus breeding were excluded from treatments 2 and 3 with electric fence. Grazed plots were stocked intensively in 2012 from April 12 through May 16. In 2013, plots were grazed from April 22 through May 6 and again from May 17 through 28. Chemical suppression was provided with 0.14 kg/ha of clethodim on March 30 in 2012 and on April 19 in 2013.

In 2012, legume seedling counts were made on June 6, or 100 DAP, and in 2013, counts were taken on June 21, or 98 DAP. Legume seedling counts were taken randomly within a rectangle 76 cm long across the two center rows about midway through the plot length.

Early spring moisture at Mound Valley in 2010 was lower than in 2011 and lower than the long-term average (Table 1). However, legume seedling density did not differ at ≥50 DAP between years, and no interaction between year and any treatment factor was detected, so two-year averages are shown. Legume species differed in seedling density, with red clover density greater (P < 0.05) than that of white clover and alfalfa (data not shown). The relative ability of red clover to establish was typical of most experiments (e.g., Guretzky et al. [14] ) owing to the larger seed size of the short-term perennial.

There was a significant (P < 0.01) interaction among the three treatments, so means within legume, pest protection, and fescue suppression treatments are shown in Table 4. Red clover seedling density was initially higher than that of the other legumes (data not shown), but by 50 DAP, seedling density of some red clover treatments was no greater than that of a corresponding treatment of birdsfoot trefoil, and even of white clover when both were untreated.

Red clover seedling density was increased by protection with insecticide only when fescue was not suppressed. Grass suppression by glyphosate increased red clover density except where insecticide was used, and suppression by clipping increased density only when no pest protection was provided. The lack of treatment additivity could indicate some interaction between the legume and effects of the pesticide(s). Seedling density of birdsfoot trefoil and white clover seemed unaffected by pest protection or tall fescue suppression. In contrast, Cuomo et al. [12] found that suppression by glyphosate of cool-season grasses enabled much greater success in obtaining legume stands that included alfalfa, birdsfoot trefoil, red clover, and kura clover (Trifolium ambiguum M. Bieb.). Seguin et al. [13] also found that sod suppression by glyphosate or physical methods contributed to higher red, and particularly white, clover densities. The limited responses to suppression in this experiment may

relate to the lack of sufficient N to invigorate tall fescue growth [14] . An overall reduction in grass competitiveness was necessary at one site for physical grass suppression methods to equal chemical suppression [13] ; thus, a recommendation for legume establishment without chemicals might be to reduce or eliminate N fertility or otherwise diminish the grass’ energy reserves in the period prior to interseeding.

The mean responses for legume seedling density to suppression treatments in 2012 and 2013 are shown by species, because legume × treatment interactions were often significant (Table 5). Year × treatment interactions were not significant (P > 0.05) for red clover or for birdsfoot trefoil. Red clover seedling densities were not significantly affected by treatment, whereas for birdsfoot trefoil, chemical suppression significantly increased seedling density compared with no treatment (the check) and grazing (Table 5).

White clover seedling density exhibited a year × treatment interaction (Table 6). This occurred because the few seedlings found in 2012 resulted in no significant treatment difference, but the higher seedling density in 2013 resulted in significant differences. The two-year average seedling density in Table 5 with grass suppression appeared significantly better than the check. Seedling densities for the treatments by year (Table 6) indicate no difference because of largely unsuccessful establishment in 2012. In 2013, however, densities from chemical and grazed treatments were similar, but both were significantly (P > 0.05) greater than the check.

Seedling densities of red clover and birdsfoot trefoil were also much higher at ~100 DAP in 2013 than in 2012 (data not shown). Poor legume establishment in 2012 may have been related to the warmer, drier conditions in May and early June compared with those in 2013 and the long-term average (Table 1), emphasizing the role of environmental conditions at critical seedling stages.