OUP user menu

Influence of Planting Date on Stink Bug Injury, Yield, Fiber Quality, and Economic Returns in Georgia Cotton

Ishakh Pulakkatu-Thodi, Donald Shurley, Michael D. Toews
DOI: http://dx.doi.org/10.1603/EC13395 646-653 First published online: 1 April 2014


Phytophagous stink bugs are economically important pests of annual and perennial crops in the southeastern United States. Because of insecticide resistance and risk of secondary pest outbreaks, there is interest in identifying cultural practices that could lead to reduced insecticide applications. The objective of this project was to assess the importance of cotton planting date on stink bug damage to cotton. Unsprayed cotton plots with biweekly planting dates were established at three locations in southern Georgia in each of two crop years. During the bloom cycle, stink bug-induced boll injury was estimated weekly in each plot. Plots were subsequently defoliated, mechanically harvested, and ginned to assess differences in fiber yield and quality attributable to stink bug injury. Results show that the rate of boll damage generally increased more rapidly through the bloom cycle for planting dates in June compared with May. Similarly, estimates of boll damage from June-planted cotton more frequently exceeded the stink bug treatment threshold compared with May-planted cotton. In 2011, mean lint yield and economic returns from May planting dates were significantly greater than June planting dates. In 2012, lint yield and economic returns were greater in plots established in early May compared with later planting dates. Estimates of HVI color + b, a measure of fiber yellowness, were lower in early May-planted cotton compared with June planting. These data show that growers need to be aware of increased stink bug damage potential when planting late.

  • cultural practice
  • integrated pest management
  • cotton fiber quality
  • Nezara viridula
  • Euschistus servus

Pest management in southeastern U.S. cotton production has shifted from traditional production systems that rely heavily on broad spectrum insecticides to integrated systems that use pest resistant cultivars and selective insecticide applications (Summy and King 1992, Greene et al. 2001). Boll weevil eradication and widespread adoption of transgenic cotton varieties targeting caterpillars are generally responsible for the significant reduction in pesticide use. However, these factors are also believed to have contributed to the emergence of the stink bug complex as an economic pest in southeastern cotton production systems (Greene and Herzog 1999, Greene et al. 2001). Preferential feeding by stink bugs on young developing cotton bolls causes abscission of young bolls or a loss of lint quality when larger bolls are damaged (Barbour et al. 1990, Willrich et al. 2004a). Of several species of stink bugs that are encountered in cotton fields, the green stink bug, Chinavia hilaris (Say) [formerly known as Acrosternum hilare (Say)], the southern green stink bug, Nezara viridula (L.), and the brown stink bug, Euschistus servus (Say), are the most common (Reay-Jones et al. 2009). Apart from the feeding damage, stink bugs are capable of transmitting cotton seed and boll rotting bacteria such as Pantoea agglomerans. Infection of P. agglomerans strain Sc 1-R can cause rotting of an entire locule (Medrano et al. 2007). Stink bugs have been consistently ranked among the most damaging insect pests of cotton in southeastern states for the past several years (Williams 2009, 2010, 2012). Approximately 0.53 million ha of cotton in Georgia were infested with stink bugs in 2011 and those infestations required insecticide treatment of ≈0.4 million ha, at an average of two applications per season (Williams 2012).

Stink bugs are reported to feed on >200 cultivated and noncultivated host species (Todd and Herzog 1980, McPherson and McPherson 2000, Reay-Jones et al. 2010). Besides cotton, they are an economic pest in row crops such as corn, Zea mays L. (Negron and Riley 1987); soybean, Glycine max (L.) (McPherson and McPherson 2000); fruit crops such as peach, Prunus persica (L.), and apple, Malus domestica (Leskey and Hogmire 2005); small grains, such as wheat, Triticum aestivum L. (Viator et al. 1983), and grain sorghum, Sorghum bicolor (L.); and (Hall and Teetes 1982) vegetables such as tomato, Lycopersicon esculentum Miller (Lye et al. 1988).

Cultural practices, such as adjusting date of planting to mitigate peak pest pressure, could be used to manage the stink bug complex. Stink bug damage is most critical during the third, fourth, and fifth week of the cotton bloom cycle. In recognition of this period, southeastern entomologists recommend the use of a dynamic treatment threshold for stink bugs whereby the treatment threshold is set lowest during this period (Greene et al. 2008, Bacheler et al 2009). In Georgia, the treatment threshold is set at 20% internal boll injury during the second week of bloom, 15% internal boll injury during the third through fifth weeks of bloom, 20% during the sixth week of bloom, and 30% during the seventh week of bloom. Manipulation of planting dates is an effective strategy to mitigate damaging populations of insect pests in other systems. For example, planting date affected abundance of stink bugs in early and late planted corn with early planted corn having significantly fewer southern green and brown stink bugs (Tillman 2010). Furthermore, uniform delayed planting was recommended in the rolling plains of Texas to manage boll weevil in cotton based on the beetle's diapause and overwintering survival (Slosser 1978). Soybean cultivars that allowed early planting in the mid-South had fewer lepidopteran defoliators but harbored more stink bugs (Baur et al. 2000).

Cotton planting in the southeastern United States generally starts in late April and continues until early June (U.S. Department of Agriculture–The National Agricultural Statistics Service [USDA–NASS] 1997). The final planting date for full insurance coverage in Georgia and Alabama is 31 May; in Florida this date is 10 June. In Georgia, coverage is reduced 1% per day for each day after 31 May up to 15 June. June-planted cotton is relatively common in Georgia because of drought or excess moisture, planting “double crop” cotton behind wheat, and in fields that require replanting as a result of cool weather, extreme moisture conditions, herbicide injury, or severe seedling insect infestations. It is desirable to identify a “safe window” of planting cotton where the risk of damage due to stink bugs could be minimized. The objective of this study was to quantify stink bug damage in terms of boll injury, yield, lint quality, and economic value in cotton planted at four different dates spanning over the typical southeastern cotton planting season.

Materials and Methods

Study Locations.

This experiment was conducted in 2011 and 2012 on experiment farms operated by the University of Georgia or USDA. In 2011, trials were conducted near Tifton (31° 30′44.159″ N, 83° 32′53.8296″ W), Midville (32° 52′20.536″ N, 82° 12′52.9704″ W), and Plains (32° 2′12.174″ N, 84° 22′2.8236″ W). Trials were repeated in 2012 near Tifton (31° 30′28.0814″ N, 83° 33′22.0129″ W) and Plains (32° 2′12.174″ N, 84° 22′2.8236″ W). A second generation cotton cultivar, “DP 0912 B2RF,” containing Cry1Ac and Cry2Ab proteins for resistance to lepidopteran caterpillars was planted both years at each location. Overhead irrigation was provided at all locations and Extension recommended agronomic practices for cotton grown on conventional tillage were followed, with the exception that no insecticides were applied after planting.

Plot Layout.

At each location, plots were arranged in a randomized complete block design with 3–5 replicates. In 2011, planting dates at Tifton were 12 May, 26 May, 9 June, and 23 June and at Midville they were 10 May, 24 May, 7 June, and 21 June. corresponding planting dates at Plains included 9 May, 23 May, 6 June, and 20 June. All plots were eight rows in width (0.91 m row spacing) and 15.24 m in length except in Midville, where the plots were 30.48 m in length. In 2012, plantings at both Tifton and Plains were conducted on 10 May, 24 May, 7 June, and 21 June. Plots at Tifton were eight rows in width and 12.19 m in length and plots in Plains were four rows in width and 15.24 m in length. Regardless of planting date or location, all plots were planted using seed from the same bag. The same pneumatic planter and planting depth was used for all plots.


A comprehensive sampling plan consisted of weekly sweep net samples for stink bugs, weekly collection of immature cotton bolls for assessing boll injury, seedcotton yield, and fiber quality assessment on a representative fiber sample from each plot. Sampling for stink bugs and bolls commenced during the second week of bloom and ended when no appropriately size bolls were available; there were no appropriately sized bolls available during the first week of bloom. In each plot, technicians used a sweep net (38.1 cm in diameter) sample of 20 sweeps from a single row that varied weekly to avoid sampling the same row in consecutive weeks. In addition, 20 (2.3–2.8 cm in diameter) soft bolls were collected randomly in each plot and no more than one boll was collected from a given plant. Consistent boll size was assured with use of a scouting decision aid comprised a stiff plastic card with two holes (2.3 and 2.8 cm); appropriate-sized bolls fit through the large hole but not through the smaller hole (Bacheler et al. 2010). Collected stink bugs and boll samples from each plot were held in labeled plastic bags for further processing in the laboratory. Boll samples were subsequently examined for internal damage, including feeding punctures, warty growths on the inner carpel wall, stained lint or rotten locules (Toews et al. 2009); stink bugs were identified to species.

At the end of the season, cotton was chemically defoliated and the center two rows from each plot were picked with a two-row spindle picker modified to collect seedcotton into individual bags by plot. The resulting seedcotton was weighed and ginned by plot at the UGA Microgin (Tifton, GA). Ginning is the process of taking seedcotton and separating the lint from the seeds. The efficiency of this process is called gin turnout, calculated by taking the lint weight divided by the seedcotton weight. Finally, representative fiber samples from each plot were sent to the USDA Classing Office located at Macon, GA, for official grading. Cotton lint classification followed USDA's official grade standards for American Upland cotton (U.S. Department of Agriculture–Agricultural Marketing Service [USDA–AMS] 2001). Lint characteristics such as color grade, leaf grade, staple length, micronaire, strength, color Rd (a measure of fiber brightness), and color +b (a measure of fiber yellowness) were determined using the Uster High Volume Instrument (HVI).

Economic analyses were conducted on lint yield per ha and resulting fiber quality based on the December 2011 and December 2012 average Georgia cash (spot) prices (USDA–AMS 2011, 2012). The average cash prices for December 2011 and December 2012 were US$1.92 and US$1.47 per kg (87.14 and 72.67 cents per pound), respectively. Baseline prices receive an incentive or discount based on the quality characteristics determined by grade standards. Fiber quality characteristics considered for analysis included color, leaf, staple (CLS), micronaire, strength, and uniformity.

Data Analysis.

Planting dates in 2011 differed by up to two calendar days with that of 2012. To avoid confusion and for the ease of analysis, those dates were synchronized with the 2012 dates. Because only a few stink bugs were captured in sweep samples, stink bug density was not analyzed beyond documentation of species composition. Percentage boll damage data were analyzed using linear regression methods because the week of bloom is a quantitative variable. Although data were collected through the seventh week of bloom in some cases, we only analyzed weeks 2 through 6 because boll availability was highly inconsistent across study sites during the seventh week of bloom. Simple linear curve models were fitted using the PROC REG procedure in SAS 9.3 (SAS Institute 2012, Cary, NC), with the week of bloom as an independent variable and the mean percentage boll injury as a dependent variable. Regression model fit was evaluated using pattern of residuals and F tests for lack of fit. Comparisons among planting dates were conducted by contrasting slopes of two planting dates at a time. The PROC GLIMMIX procedure with LSMEANS statement was used to compare seedcotton yield, lint yield, cotton fiber quality parameters, and economic returns among the four planting dates. Data across locations were pooled together for analysis. In 2012 only, yield data from the two June planting dates at Tifton were omitted from the analysis because of poor stand establishment.


Boll Damage.

In general, there was less boll damage observed in the early planting dates compared with later planting dates. Mean percentage boll damage due to stink bug feeding in plots planted on four different planting dates (10 May, 24 May, 7 June, and 21 June) in 2011 was 6.5 ± 1.0, 9.7 ± 1.2, 15.6 ± 1.9, and 18.8 ± 2.4, respectively. Regardless of the planting date, boll damage generally increased with successive weeks of bloom. The earliest sampling date in 2011 was 14 July and in 2012 was 16 July. In 2012, for the respective planting dates, mean percentage boll damage was 11.8 ± 1.6, 13.6 ± 2.1, 22.8 ± 3.6, and 21.7 ± 3.9. The overall mean percentage boll damage was numerically greater in 2012 (17.3 ± 1.5) compared with 2011 (12.6 ± 0.9).

Linear increases in percentage boll damage allowed fitting a regression line through the temporal data and comparison of linear slopes among the four planting dates. In 2011, regression lines for different planting date were described by the equations y = 2.3(±1.0)x − 2.5(±4.3) for 10 May, y = 3.5(±1.0)x − 4.0(±4.3) for 24 May, y = 6.0(±1.0)x − 8.6(±4.4) for 7 June, and y = 7.5(±1.0)x − 10.9(±4.2) for 21 June, where y is mean percentage boll damage and x is week of bloom. The overall model for 2011 showed significant differences in slopes among planting dates (F = 56.43; df = 8, 240; P < 0.01) with an adjusted r2 value of 0.64. Statistical comparison of the linear slopes showed a significant increase in the amount of damage in the June planting dates in 2011 (Fig. 1). corresponding regression lines in 2012 were y = 2.5(±1.8)x − 2.0(±7.7) for 10 May y = 5.4(±1.8)x − 8.1(±7.6) for 24 May, y = 7.2(±2.0)x − 6.6(±8.3) for 7 June, and y = 8.0(±1.9)x − 9.3(±7.8) for 21 June. Only the late June planting date exhibited a significantly greater slope than the remaining treatments (F = 29.61; df = 8, 126; P < 0.01; and adjusted r2 = 0.63; Fig. 1). Examination of actual calendar dates, as opposed to the week of bloom, shows that the mean percentage boll damage increased during the first half of August each year (Table 1).

Fig. 1

The mean percentage boll damage by the week of bloom for four cotton planting dates in 2011 (top) and 2012 (bottom). Comparisons among planting dates were conducted by contrasting the slopes of two planting dates at a time. Statistically different slopes are indicated by different letters (P < 0.05).

View this table:
Table 1

Mean ± SEM percentage boll injury across planting dates by year, month, and calendar week


The percentage boll injury in plots with June planting dates exceeded the Extension recommended treatment threshold for stink bugs much more frequently than those with May planting dates. In 2011, the percentage boll injury for plots in both the 10 and 24 May planting dates never exceeded the threshold. However, damage to bolls from both June planting dates exceeded the threshold on three of the five possible dates. In 2012, the percentage boll injury in the 10 May planting date exceeded the threshold in the fifth week of bloom only, while the 24 May plating date exceeded the threshold in both the fifth and sixth weeks of bloom. In contrast, plots in both 7 and 21 June planting dates exceeded the treatment threshold in the third, fourth, fifth, and sixth weeks of bloom.

Yield and Gin Turnout.

The mean seedcotton yield (kg/ha) and mean lint yield (kg/ha) differed significantly as a function of planting date in 2011 (F = 18.35; df = 3, 36; P < 0.01; and F = 23.43; df = 3, 36; P < 0.01). In 2011, both planting dates in May exhibited a statistically similar seedcotton yield, which was significantly greater than the yield from both June planting dates (Table 2). Analysis of mean lint yield showed that cotton planted in early and late June yielded significantly lower lint compared with cotton planted in May (Fig. 2), with late June-planted cotton yielding the least lint. Percentage gin turnout was marginally nonsignificant among planting dates (F = 2.75; df = 3, 36; P = 0.06). In 2012, mean seedcotton yield (kg/ha) and mean lint yield was significantly greater for the 10 May planting date (F = 8.20–7.56; df = 3, 10; P < 0.01; Table 2; Fig. 2). The percentage gin turnout was significantly greater for both May planting dates compared with June planting dates (F = 13.08; df = 3, 18; P < 0.01).

View this table:
Table 2

Mean ± SEM of various parameters evaluated for cotton planted at four different planting dates in Georgia in 2011 and 2012

Fig. 2

The mean lint yield (kg/ha) recorded for four different planting dates in 2011 (top) and 2012 (bottom). Means within a year followed by different letters are significantly different. LSMEANS test (P < 0.05).

Fiber Quality.

In 2011, the planting date significantly affected fiber yellowness and brightness as indicated by the HVI color +b value (F = 68.23; df = 3, 36; P < 0.01) and HVI color Rd value (F = 19.18; df = 3, 36; P < 0.01). The HVI color +b value and HVI color Rd value together determine color grade of the cotton fiber. The May planting dates exhibited statistically comparable color +b values, which were significantly less yellow than the June-planted cotton (Fig. 3). Conversely, HVI color Rd value indicated brighter lint in the late June planting date. Planting date was not a factor in defining variability in staple length (F = 2.19; df = 3, 36; P = 0.10); however, planting date was a factor in determining the variability in staple strength (F = 4.78; df = 3, 36; P < 0.01) and uniformity (F = 5.16; df = 3, 36; P < 0.01). Interestingly, these values indicated better quality for late planted cotton.

Fig. 3

The mean HVI color +b value recorded for lint harvested from four different planting dates in 2011 (top) and 2012 (bottom). Means within year followed by different letters are significantly different. LSMEANS test (P < 0.05).

Fiber samples were also available from all sites in 2012. HVI color +b value was significantly better for 10 May-planted cotton (F = 11.79; df = 3, 18; P < 0.01), while cotton planted on remaining planting dates had comparable values (Fig. 3). Fiber uniformity was significantly affected by planting date (F = 4.01; df = 3, 18; P = 0.02) and indicated better values toward late planting dates. Other quality parameters that showed significant variability were staple length (F = 6.60; df = 3, 18; P < 0.01) and strength (F = 6.19; df = 3, 24; P < 0.01); in both cases the late June planting date was different from the remaining planting dates. HVI color Rd (F = 2.74; df = 3, 18; P = 0.07) did not vary significantly based on planting date in 2012.

Stink Bug Capture.

The number of stink bugs captured in sweep net samples was generally very low in both years. In 2011, 287 samples (20 sweeps per sample) were conducted and only 14 stink bugs were captured. Of these, 42.8% were brown stink bug and 57.1% were green stink bugs; no southern green stink bugs were captured. Much greater stink bug pressure was observed in 2012. From 166 sweep net observations, 39 stink bugs were captured, of which 92.3% were southern green stink bugs and the rest were brown stink bugs.

Economic Analysis.

Lint value based on fiber yield and cash prices (December 2011 and December 2012), adjusted for fiber quality, differed significantly by planting date in 2011 (F = 21.33; df = 3, 36; P < 0.01) and 2012 (F = 6.27; df = 3, 10; P = 0.01). Both May planting dates had statistically similar lint values, which was significantly greater than the June planting dates in 2011. Late June-planted cotton produced the least lint value (Fig. 4). Early May-planted cotton had significantly greater lint value in 2012 (Fig. 4) compared with the remaining planting dates. Lint value was primarily driven by the lint yield as comparison of adjusted base price did not indicate significant differences by planning date (2011: F = 1.93; df = 3, 36; P = 0.14; 2012: F = 1.93; df = 3, 18; P = 0.19).

Fig. 4

Mean lint value (US$/ha) by planting date for 2011 (top) and 2012 (bottom). Scale bars denoted by same letters are not significantly different. Means within a year followed by different letters are significantly different. LSMEANS test (P < 0.05).


These studies demonstrated that June-planted cotton was generally at a higher risk of being damaged by stink bug populations compared with May-planted cotton in Georgia. The mean percentage boll damage was consistently high and exceeded the treatment threshold more frequently in the later planted cotton plots. The mean percentage boll damage by calendar weeks showed that the numerically greatest percentage damage (42.9%) recorded was during the third week of September in 2011 and the second week of September in 2012 (38.0%; Table 1). On the contrary, the mean percentage boll damage was generally low (<10%) from early July through early August in both years. Use of internal cotton boll damage as an accurate estimation for the presence of stink bugs and the correlation with other sampling methods has been established in previous research (Toews et al. 2008). Our study implies that planting cotton early in the planting window will allow growers to escape peak stink bug pressure and thereby possibly eliminate or minimally reduce the number of sprays required to manage them.

The optimal planting window for cotton is when the soil temperature at the 5-cm depth approaches 18°C and the weather forecast is favorable for crop development (Silvertooth et al. 1999). The typical planting season in Georgia usually lasts from 20 April to 5 June (USDA–NASS 1997). However, Georgia growers plant cotton much later (up to 20 June), after a spring crop or when the entire field must be replanted due to a wide variety of biological (insect, nematode, or plant disease) and abiotic (extreme drought or rainfall events, herbicide injury, or sandblasting) factors. Early planting in general is thought to improve yield (Pettigrew 2002, Boquet and Clawson 2009), but moderately late planting, i.e., 10–15 d after the recommended planting window, does not always reduce cotton yield (Slosser 1993, Bauer 1998). Of the four planting dates selected, the first two planting dates were well within the typical planting window in Georgia. The third planting date was timed to occur near the end of the typical planting window and the fourth planting date would only occur in replant or double crop situations. In our study, yield reduction was consistent in late planted cotton. The potential of stink bugs to reduce yield in cotton has been demonstrated previously (Barbour et al. 1990, Willrich et al. 2004b). Based on the increased boll injury observed in late planted cotton, there is reason to believe that stink bugs contributed to the observed yield loss in this study. Lint yield was reduced up to 36% in 2011 in late June-planted cotton when compared with lint yield of early May-planted cotton.

Stink bug feeding on developing cotton bolls can influence fiber quality in several ways. Feeding affects physical properties of cotton fiber, such as lint color, micronaire (a measure of fiber maturity), uniformity, staple length, and strength (Cassidy and Barber 1939, Barbour et al. 1990, Bommireddy 2007). Toews and Shurley (2009) showed that quality of cotton fibers harvested from edges of the cotton field were negatively affected because of higher concentration of bugs in borders shared with other crops. Here, HVI color +b values were lower in both May planting dates in 2011 and in 10 May-planted cotton in 2012. Fiber reflectance, measured by HVI color Rd values, indicated better fiber quality in later planted cotton during 2011, but this was not evident in 2012. Other quality parameters, such as staple length, strength, and uniformity, indicated better quality in June-planted cotton in at least 1 yr. Here, the influence of stink bug damage was not evident in quality parameters except HVI color +b values. Along with genetics, various environmental factors, primarily weather, are known to affect cotton fiber quality (Bradow 2000). It is also possible that changes in weather factors due to delayed planting masked the influence of stink bug on these quality parameters.

Economic returns were consistently greater in the early May-planted cotton. Statistically, lint yield was the most important factor that influenced economic returns, as returns followed nearly the exact same trend as those exhibited by lint yield (Fig. 2 and Fig. 4). Previous research (Toews and Shurley 2009) showed that stink bug damage can affect the economic value of lint. Although there were observed differences among planting dates in some of the other fiber quality parameters measured, those subtle differences were not sufficient to affect economic returns. Differences in HVI color Rd may reflect changing environmental conditions, such as rainfall, after the boll opened. Similarly, HVI strength and length vary as a function of weather when the fibers are developing and likely have nothing to do with stink bug feeding. Considering that the optimal planting window starts in late April, there may be potential for further improvement in yield and fiber quality by planting earlier than 10 May.

Low stink bug capture in the sweep net does not necessarily suggest less stink bug pressure. An observed lack of stink bug sampling efficiency using sweep nets is well documented; conversely, assessment of internal boll feeding injury has been shown to be much more sensitive to changes in stink bug density in the field (Toews et al. 2008, Reay-Jones et al. 2009). Nevertheless, species composition would not be possible if only boll damage were assessed. Here, gradual increases in percentage boll damage with characteristic stink bug feeding symptoms established that a greater density of stink bugs was present in the crop later in the season.

Data presented here strongly suggest that planting early is a good way to escape stink bug injury. The authors suggest that there are two explanations for decreased boll damage in early planted cotton. First, the earliest planted cotton started blooming in early July when there would be many other suitable stink bug hosts, both agronomic and wild, in the farmscape. Conversely, the 21 June-planted cotton did not start blooming until mid-August when some wild hosts and agronomic hosts like corn would dry to the point of no longer being attractive to stink bugs. Blooming cotton may attract a larger percentage of the stink bug population during August and September. Second, southern green stink bugs are multivoltine and there would be time for an additional generation to develop by the time the June-planted cotton was most attractive.

A better understanding of stink bug population dynamics, movement, and alternate host usage will contribute to improved cultural control practices for stink bugs. Previous research by Carrière et al. (2012) demonstrated that landscape factors, such as adjacent habitats of alternate hosts of an insect pest, can influence population dynamics of the pest in cotton fields. Toews and Shurley (2009) observed that cotton grown in close proximity to specific agronomic hosts, for example soybean and peanut, was deleteriously affected by stink bug feeding as quantified by seedcotton yield, gin turnout, lint color, and lint value. Tillman (2006) showed that sorghum could be used as an effective trap crop to protect cotton from stink bugs. To this growing list of cultural practices for managing stink bugs, the authors submit that early planting is an equally import tool.


We gratefully acknowledge Annie Horak, David Griffin, and Jerry Davis for their excellent technical support. Miguel Soria, Jamal Hunter, Ta-I Huang, and Barry Luke assisted in field scouting and processing cotton bolls at the laboratory and we acknowledge their efforts. This project was funded by U.S. Department of Agriculture–Cooperative State Research, Education, and Extension Service–Current Research Information System–Group on Earth Observations (USDA–CSREES–CRIS–GEO; 0062) and USDA–CSREES Special Research Grant program award 2010-34566-21116.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions{at}oup.com

References Cited

View Abstract