Proceedings of the Third International Conference on Invasive ...

Chapter 4: Spartina Control and Management<strong>Proceedings</strong> <strong>of</strong> <strong>the</strong> <strong>Third</strong> <strong>International</strong> <strong>Conference</strong> on Invasive Spartinacrushing 2 yrscrushing + glyphosate38 kg/ha 1 yrglyphosate 9 kg/ha 1 yr0 20 40 60 80 100% Spartina-free quadratscrushing 1 yrcrushing +glyphosate38 kg/ha 2 yrsglyphosate 2 yrsglyphosate 38 kg/ha 1 yrFig. 1 Box whisker graphs <strong>of</strong> percent Spartina-free quadrats from differentmechanical and chemical controls used in Willapa Bay from 2001 to 2003.Box whisker graphs only presented for treatments with five or more datapoints. (mean = dotted line, median = solid line, box = 25 th and 75 thpercentiles, whiskers = 5 th and 95 th percentiles, data point = outliers).Three mechanical control practices (tilling, crushing anddisking) and four chemical control practices (glyphosate attwo rates, crush and spray, and imazapyr) were assessed.Tilling was done by <strong>the</strong> WNWR in <strong>the</strong> winter on largemeadows that were previously mowed (2000/2001) oruntreated (2004). Crushing was done by WSDA, WDNR orWDFW using tracked amphibious vehicles. Disking wasdone with a tandem disk. Crushing timing and frequencyvaried by site while disking was done only in <strong>the</strong> winter.Sites usually received multiple crushing or disking eventsper year. Imazapyr and <strong>the</strong> low rate <strong>of</strong> glyphosate (9kilograms per hectare (kg/ha)) were applied by boom usingapproximately 500 liters per hectare (l/ha) spray volume.The high glyphosate rate was applied by hand using highpressure spray guns. A 5 to 8% v/v <strong>of</strong> product was appliedwith a spray volume <strong>of</strong> about 1,500 l/ha. The estimatedglyphosate rate for hand applications averaged 38 kg/ha.Spraying <strong>of</strong> crushed sites occurred once Spartina had regrownenough to provide adequate canopy coverage.Treatment sites ranged from 0.25 ha to 500 ha in size with<strong>the</strong> majority larger than 10 ha.Attempts were made to assure that <strong>the</strong> treatment sitesselected for monitoring were relatively free <strong>of</strong> confoundingfactors from previous control efforts at <strong>the</strong> sites, and thateach site had a discrete treatment that was identifiable.Monitoring was done along multiple transects in eachtreatment site. Spartina stem and seedling density countsfrom 0.25m 2 quadrats were taken along transects. Thenumber <strong>of</strong> quadrats and type <strong>of</strong> transect varied depending on<strong>the</strong> size and shape <strong>of</strong> each treatment area, with a range <strong>of</strong> 40to 2,600 quadrats per site. Methods are fully detailedelsewhere (Patten 2003, 2004). Stem density data wereclassified as follows:1) excellent Spartina control – 0 stems/0.25m 2 ;2) good to moderate control – 1 to 5 stems/0.25m 2 ;3) fair to poor control – 6 to 20 stems/0.25m 2 , and4) poor to no control – 21 stems/0.25m 2 . Mean stemdensity before treatment was typically >50 stems/0.25m 2 .Pooled and disaggregated data for each site were analyzedfor stem density and overall stem density frequencydistribution. Means and standard error values are presented.The percent <strong>of</strong> Spartina-free quadrats was used as aconservative method to assess treatment efficacy. This valueindicates a treatment’s ability to prevent vegetative recolonizationand to minimize <strong>the</strong> cost <strong>of</strong> re-treatment. Forexample, a thinned-out Spartina canopy with a mean density<strong>of</strong> 5 stems/0.25m 2 could represent a greater than 90%decrease in stem density, but would cost almost <strong>the</strong> same totreat as a solid canopy; without re-treatment this wouldbecome a solid infestation again within a year. For this paper,<strong>the</strong> percent <strong>of</strong> Spartina-free quadrats was also assumed toapproximate <strong>the</strong> extinction coefficient or <strong>the</strong> percent control.Years to achieve eradication for a given treatment, assumingminor re-infestation, was calculated as years = log 0.01/log (1-extinction coefficient). A discrete time, logistic growth model,with growth rate and outside seedling input parameters, wasused to predict <strong>the</strong> amount <strong>of</strong> Spartina meadow remainingafter four years <strong>of</strong> a given treatment.Parametric comparisons <strong>of</strong> treatment efficacy were notfeasible because treatments lacked true replication. That is,even sites with <strong>the</strong> same general treatment were treated bydifferent agencies under vastly different conditions.Therefore to make inferences about overall treatmenteffectiveness from a management perspective, treatmentvariation is presented in box whisker format (mean, median,5 th , 25 th , 75 th and 95 th percentiles, and outliers) for alltreatments where <strong>the</strong>re were more than five data sets.For comparative purposes, box whisker graphs <strong>of</strong>efficacy on Spartina across selected herbicides are shown;this research was conducted by <strong>the</strong> author between 1998 and2003 (Patten 2002; Patten & Stenvall 2002). Percent controldata were pooled across each experimental unit forglyphosate at 18 kg/ha rate and imazapyr at 1.7 kg/ha.RESULTSIndividual and pooled results are shown in Tables 1 and2, and Fig. 1. Control with <strong>the</strong> broadcast rate <strong>of</strong> glyphosate(9 kg/ha) was highly variable (14% to 56%) and averagedhalf <strong>the</strong> efficacy achieved at <strong>the</strong> much higher hand-sprayedrates (38 kg/ha) (Table 1 and Fig. 1). There was less- 250 -