Conditions to test electric fence additions to cattle barb wire fence for goat containment

Abstract

Two experiments were conducted to determine appropriateness of conditions in a method being developed for evaluating efficacy of different electric fence additions to cattle barb wire fence for goat containment. In Experiment 1, two 6×6 Latin squares (LS), each with 24 yearling Boer goat doelings previously exposed to electric fence, were conducted. After overnight fasting, groups of four doelings were placed in 2.4×2.4 m pens without forage. One pen side was five strands of four-point barb wire (non-electrified) at 31, 56, 81, 107 and 132 cm from the ground adjacent to a pasture with abundant vegetation. One LS had periods 2–3 days in length and the other 7 days. Electric fence treatments for each square were addition to barb wire fence of four electric fence strands 15, 28, 43 and 58 cm from the ground at low voltage of 4–4.5 kV (4S-LV); two strands at 15 and 43 cm and high voltage of 8.5–9 kV (2S-HV); two strands at 15 and 43 cm and low voltage (2S-LV); one strand at 15 cm and low voltage (1S-LH-LV); one strand at 43 cm and low voltage (1S-HH-LV) and one strand at 23 cm and high voltage (1S-MH-HV). Percentages of doelings exiting (6% and 4%) and shocked in 2 h (15% and 16% for 7 and 2–3 days, respectively) were low and did not differ between period lengths. The percentage of doelings exiting in 2 h was not affected by fence treatment. Period of squares affected (p<0.05) the percentage of doelings shocked (54%, 25%, 4%, 6%, 0% and 4% for periods 1, 2, 3, 4, 5 and 6, respectively). Experiment 2 was with 30 Boer and 30 Spanish growing doelings in the same study area. Because of less than anticipated shock and exit in Experiment 1, some conditions were changed, including a defined period of exposure to electric fence, training for pen exit before the experiment and longer fasting (24 or 36 h). Fence treatments were those of Experiment 1 but without 4S-LV and with slightly lower voltage. Doelings were divided into three sets of 20 and used in a completely randomised design (CRD), and one set continued repeated exposure to the different fence treatments in a 5×5 LS. Thereafter, period 1 was repeated in period 6. For the CRD approach, the percentage of doelings exiting in 1 h was >90%. With the LS method the percentage of doelings exiting also was similar among fence treatments but was 75%, 70%, 40%, 70% and 75% for 2S-HV, 2S-LV, 1S-LH-LV, 1S-HH-LV and 1S-MH-HV, respectively. With a comparison involving doeling sets used in the LS, the percentage of doelings shocked was lower (p<0.05) in period 6 vs. 1 (5% vs. 50%), although there was no difference with doelings not used in the LS. In conclusion, results were not promising for successful use of an LS approach, and large differences between experiments in levels of shock and exit indicate need for further change in conditions.

Keywords:

• goat
• fence
• containment

Introduction

Many cattle producers have interest in co-grazing goats to achieve profit from goats, increase total animal production per unit land area with low dietary overlap between species (Animut and Goetsch 2008), improve vegetation conditions for cattle grazing and lessen or avert costs of other means of vegetation management (Hart 2001). Moreover, because cattle and goats do not share the same internal parasites, their co-grazing at appropriate stocking rates can decrease need for use of commercial anthelmintics, for which resistance by internal parasites is increasing (Sahlu et al. 2009).

One constraint to co-grazing of cattle and goats in many parts of the world is the different types of fence required for containment. Therefore, knowledge of effective means of modifying existent cattle barb (i.e. barbed) wire fence for goat containment, which vary in cost and labour of installation and maintenance, would be beneficial. The most common method employed is to add electric fence strands to cattle barb wire fence, with the number of barb wire strands typically five but ranging from three to six. From interaction with livestock producers and electric fence supply companies, relevant information available regarding such modifications is largely anecdotal. In this regard, there is need for development of an accurate and repeatable means of evaluating, rating, ranking or comparing efficacy of different additions of electric fence strands to cattle barb wire fence for goat containment. Examples of potential areas of study that would be facilitated are effects and interactions of different types of commercial electric fence strands, number of electric fence and barb wire strands, type of barb wire strands (e.g. number of points), height and horizontal positions of electric fence strands, electric fence strand voltage, small ruminant species, genotype, age, size and gender, etc. Objectives of this study were to determine appropriateness of conditions in a method being developed for evaluating efficacy of electric fence additions to cattle barb wire fence for goat containment. Specific conditions given attention in these experiments were length of periods of Latin squares (LS) and use of different experimental designs.

Materials and methods

Protocols for both experiments were approved by the Langston University Animal Care and Use Committee.

Experiment 1

Study area

The study area was located at one end of a 0.4-ha pasture with abundant grasses, forbs and mimosa (Albizia julibrisin) trees. Six 243×243 cm evaluation pens were situated adjacent to one another. One side of the pens had a five-strand barb wire fence. Barb wire was Style No. 33 of Sheffield Wire Products (The Burly Corporation of North America, Burleson, TX, USA), which is 12.5 gauge (3 mm²) with 14 gauge (2.0 mm²) four-point barbs at 12.7 cm spacings. Barbs are wrapped around one of the two main strands and interlocked with the second strand. Barb wire strands were 30.5, 55.9, 81.3, 107.7 and 132.1 cm from the ground. Strands were attached to steel posts located at the pen corners every 243 cm and a post in the centre of the pen in the fence line (121.5 cm from pen sides). The other three sides of the pens were welded wire mesh panels (i.e. 16.2×20.3 cm openings; four gauge, 19 mm²), with a portion of the side opposite the barb wire fence used as a gate. Six other pens (i.e. fasting) were located adjacent to evaluation pens. Vegetation in both fasting and evaluation pens was removed initially by mowing, followed by use of a hand-held device with an engine-powered rotating nylon cord (i.e. ‘string trimmer’).

Treatments

Forty-eight yearling Boer doelings [initial age and body weight (BW) of 1.5±0.02 yr and 39.0±0.84 kg, respectively] were used. These animals had been previously employed in a nutrition experiment in a confinement facility. Doelings were first allocated based on BW to two sets of 24 to be used in the two 6×6 LS. The sets were then divided into six groups, each consisting of four doelings. The assignment entailed ranking based on BW from greatest to least and random assignment to fence treatments within groups of six to minimise variability within group. LS were constructed so that groups exposed to particular treatments in one period were not subjected to one of the other treatments in preceding periods more than twice. Doelings had been previously exposed to electric fence in various pastures of the farm of the American Institute for Goat Research of Langston University. The doeling sets were taken directly from the two pastures where they resided to the study area when the fasting period began before each measurement period. These two ‘home’ pastures did not include electric fence.

Before measurements, doelings were fasted overnight and without water in the fasting pen for approximately 12 h. The purpose of fasting was to create an impetus for doelings to exit evaluation pens into the area with vegetation available for consumption. The six periods of the LS were 7 or 2–3 days in length. A period was defined as the time between measurements. Different period lengths were evaluated in regards to desirability for short periods to minimise trial length and maximise the potential number of experiments that might be conducted in a given period of time.

Different electric fence treatments were used in the six evaluation pens, the intent of which was to have varying degrees of prevention or hindrance of exit from pens, initially anticipated to range from 0% to 100%. Thus, study of these specific electric fence treatments was not an objective but, rather, the treatments were employed to evaluate effects of length of LS periods. Electric fence (without barbs) was situated 12.7 cm from barb wire strands inside pens, held in place by insulators connected to steel posts located at the pen sides and also by a 12.7-cm ‘offset’ insulator attached to the steel post in the centre of the barb wire fence line. Fence treatments were four electric strands at low voltage (4S-LV); two strands at high voltage (2S-HV); two strands at low voltage (2S-LV); one strand, low height and low voltage (1S-LH-LV); one strand, high height and low voltage (1S-HH-LV) and one strand, medium height and high voltage (1S-MH-HV). Line height from the ground was 15.2, 27.9, 43.2 and 58.4 cm for 4S-LV; 15.2 and 43.2 cm for 2S-HV and 2S-LV; 15.2 cm for 1S-LH-LV; 43.2 cm for 1S-HH-LV and 22.9 cm for 1S-MH-HV. Electric fence strands were 14 gauge (2.0 mm²) XL aluminum wire of Gallagher USA (North Kansas City, MO, USA). Aluminum wire was used because, according to the manufacturer, it conducts electricity four times better than steel wire, is one-third the weight of steel wire, is relatively flexible and will not rust. Two electric fence chargers or energisers, with grounding terminals connected to three ground rods placed 2.8 m deep and 2.8 m apart, were used for the different voltages. An electric strand was attached to strands with each voltage, with ground contact varied intentionally to achieve 4.0–4.5 and 9.5–9.0 kV for LV and HV treatments, respectively. Voltage was checked at the beginning, middle and end of measurement periods. Based on anecdotal evidence, it was anticipated that pen exit would be lowest for 4S-LV and 2S-HV, greatest for 1S-HH-LV and intermediate for other fence treatments.

Measures

Doelings were continuously observed by three individuals (two groups per observer) for 2 h after entering the evaluation pen from the pen for fasting. Variables assessed included the occurrence and number of shocks per doeling, exit from the pen (percentage of total doelings), times of shock and exit after entering evaluation pens (in 15-sec intervals) and reaction to shock (1–5, with 1 = mild and 5 = intense). A shock was recorded when an animal exhibited a noticeable reaction upon contact with an electric fence strand. However, there were no effects of treatments on reaction in this or the subsequent experiment. Also, there were no instances in which a doeling exited a pen without receiving a shock. After the 2-h observation period, observers left the area for 4 h, after which time any doelings that had exited during that time were recorded.

There was a preliminary period before period 1 of the LS. In this preliminary period, a barb wire only treatment was included rather than 4S-LV. The group exposed to this treatment was that subjected to the 1S-LH-LV treatment in the subsequent period (i.e. period 1). Impact of this prior exposure was not readily apparent based on results addressed later. All animals quickly exited from the barb wire only pen in the preliminary period, questioning rationale for its inclusion. Moreover, pen exit from other pens was greater than in the subsequent six periods of the LS. These factors suggested need to substitute a fence treatment that presumably would be very prohibitive to pen exit (i.e. 4S-LV). However, in this preliminary period the centre steel post in evaluation pens to which barb wire and electric fence strands were attached during the subsequent periods was absent. It was felt that inclusion of that centre post would be beneficial to inhibit pen exit by minimising strand movement upon animal contact. Although, as noted later, results of Experiment 1 were not supportive of this change, resulting in centre post removal for Experiment 2.

Statistical analysis

Average group values in each period were analysed with the MIXED model procedure of the Statistical Analysis System (Littell et al. 1996). First, normality was evaluated by the Shapiro–Wilk statistic of the Univariate procedure (SAS 1990). Variables except for time of the first shock were not normally distributed. Various transformations (log, square root, arc sine, inverse) were evaluated. Some transformations were normally distributed, but there was not one transformation that did so with all variables and not all variables could be made normal by any transformation. Reasons for this entail the unexpected results to be addressed later, which is relevant to the second experiment as well. Therefore, for purposes of this and the subsequent experiment to identify appropriate and inappropriate conditions for development of a study model, untransformed data were felt acceptable and used for analysis. In support, Snedecor and Cochran (1978) stated ‘many results that are useful in statistical work, although strictly true only when the population is normal, hold well enough for rough-and-ready use when samples come from non-normal populations’. Furthermore, Bradley (1978) in his treatise on robustness cites various statistical studies supporting that the t-test and F-test are extremely robust and fairly immune to deviations from normality.

The model included period length (i.e. LS), fence treatment, period length×fence treatment and period of LS. Group within period length was the random variable used to test the effect of period length; period was a repeated measure and group within period length was the subject. Seven covariance structures were evaluated with the Schwartz Bayesian criterion (Ante-dependence; Autoregressive, AR(1); Compound Symmetry; Heterogeneous Compound Symmetry, CSH; Huynh–Feldt; Unstructured; Variance Components, VC). The number of shocks and percentage of doelings receiving a first shock were analysed with CSH, AR(1) was used for the percentage of doelings receiving a second shock, and VC was employed for time of a first shock and percentage of doelings exiting in 2 and 6 h. Satterthwaite approximate was used to calculate denominator degrees of freedom. The lsmeans were estimated, and means separation was by Least Significant Difference when overall effects were significant (i.e. protected F-test; p < 0.05).

Experiment 2

Modification of procedures

Many procedures were the same as in Experiment 1, although because of findings of that experiment some modifications were made. Thirty growing Boer (initial age and BW of 0.41±0.009 yr and 20.9±0.68 kg, respectively) and 30 Spanish doelings (initial age and BW of 0.43±0.010 yr and 17.3±0.42 kg, respectively), initially without prior exposure to electric fence, were used. Younger animals were employed than in Experiment 1 because of greater expected conduciveness of their small size for pen exit under or between wires. Different breeds were employed as a consequence of anecdotal evidence suggesting greater propensity for pen exit by Spanish vs. Boer goats. These animals had been previously used in a supplementation experiment on pasture while suckling. Vegetation was initially removed from fasting and evaluation pens with herbicide and later as in Experiment 1.

For 1 month before Experiment 2, doelings used plus several additional ones of each genotype were placed in two pastures and supplemented with concentrate. Concentrate was placed daily in feeders near a strand of electric fence in each pasture with a moderate voltage of 5–6 kV. Throughout this period, all doelings consumed concentrate and numerous shocks were observed; it was assumed that all animals received one or more shocks. Next, after fasting for approximately 36 h, doelings were placed in groups of four in an evaluation pen with barb wire and no electric fence strands. All doelings quickly exited the pen into the pasture area with abundant vegetation where they were allowed to remain for several hours. One week later, the same procedure was repeated, but with one strand of electric fence included in the evaluation pen at a height of 22.9 cm at 0 kV. Again, all doelings exited the pen, most immediately upon entry. A small number took slightly longer to exit (>1 min) and were excluded from the experiment to yield 30 of each genotype. These procedures were undertaken to ensure similar adaptation to electric fence and that doelings became accustomed to potential exit from evaluation pens and subsequent vegetation consumption. After the preliminary period, doelings were only exposed to electric fence strands during measurements.

Another change in conditions from Experiment 1 involved attachment of barb wire strands, as well as electric strands, to the steel post in the centre of the evaluation pens. At least with the yearlings of Experiment 1, it was felt that the degree to which this restricted movement of strands upon contact could have been too prohibitive to exit. Hence, the post was removed.

Based on anecdotal observations, in Experiment 1 it seemed that with the HV treatments, upon shock doelings tended to jump forward more than did doelings on LV treatments, for which shock appeared to elicit somewhat more backward movement. Hence, voltages were decreased slightly for Experiment 2. Relatedly, maintaining high voltage in practical production settings is difficult with long fence distance and vegetation grounding and requires a powerful and expensive energiser. In Experiment 2, rainfall was less than in Experiment 1. Therefore, the ground under and near electric fence strands was wetted a few minutes before measurement periods began.

As noted for the adaptation procedures, length of fasting was initially increased to 36 h. This alteration was also to increase the desire to exit evaluation pens. However, because this experiment and also Experiment 1 were conducted in the autumn, with the onset of cooler nighttime temperature in latter periods, length of fasting was decreased to 24 h. Likewise, with the longer period of fasting, doelings were housed during that time in a pen in an enclosed facility that had a 6.1×1.35 m area with a concrete floor and a 6.1×4.25 m unpaved floor area with bedding of pine shavings and water available free-choice. Doelings were transported to the study area on the morning of observation.

Because of low pen exit in Experiment 1 regardless of fence treatment, particularly in latter periods, the 4S-LV fence treatment was excluded from Experiment 2. This change also related to the need to minimise the number of treatments so as to maximise the number of observations, probably most important for the completely randomised design (CRD) approach of period 1 necessitating a larger number of animals compared with the LS design.

Treatments and measures

The five fence treatments were 2S-HV, 2S-LV, 1S-LH-LV, 1S-HH-LV and 1S-MH-HV of Experiment 1, but with voltage of 3.5–4.0 and 7–8 kV for LV and HV treatments, respectively. The 30 Boer and 30 Spanish doelings were allocated to three groups of 10 per genotype for similar mean BW and variation in BW within groups. Groups of 10 doelings of each genotype were combined to form three doeling sets of 20. Sets were assigned to five groups per set, with two doelings of each genotype per group. In contrast to Experiment 1, assignments were made to achieve similar BW and variation in BW within groups. This was performed because of the CRD approach, which entailed exposure of groups to only one fence treatment rather than to all treatments as with a LS.

The entire experiment entailed six periods. In period 1, on sequential days the three doeling sets were used for measures, with one group of four doelings (two per genotype) present in each evaluation pen and exposed to the different fence treatments. Thus, period 1 was a CRD. Then, doelings of set 2 were exposed to the other fence treatments each week in periods 2–5 for a 5×5 LS. The assignment of groups of doelings to fence treatments in the five periods was as in Experiment 1. After the last period of the LS, period 1 was repeated in period 6.

As a consequence of considerably different results of Experiment 2 than those of Experiment 1, continuous observation occurred for a 1-h period, with recording of any doelings that exited pens in the subsequent 1-h period. Measures were generally the same as in Experiment 1. However, in Experiment 2 there was considerable pen exit without noticeable shock, which was also recorded.

Statistical analysis

Data were averaged by group in each period and analysed in five methods with MIXED models. The first method was with period 1 data as a CRD; the model consisted of fence treatment, breed, fence treatment×breed and doeling set. Doeling set was the error term to test the effect of fence treatment, and breed within set was a random effect as well.

The second method of analysis was for doeling set 2 as a LS in periods 1–5. The model included fence treatment, period, breed and fence treatment×breed. Doeling group was the random variable used to test effects of fence treatment and period, breed within group was used to test the effect of breed, and period was a repeated measure with a subject of breed within group.

The third analysis was a comparison of all period 1 data as a CRD with that of the LS with doelings of set 2, abbreviated as CRD-LS. The model included method (i.e. CRD vs. LS), fence treatment, method×fence treatment, breed, fence treatment×breed and method×breed, with random effects of group and breed within group, and the effect of period in the LS was not addressed. The three-way interaction was non-significant for all variables.

In the fourth analysis, data of sets 1 and 3 in periods 1 and 6 were analysed together, abbreviated as Period 1&6-set 1&3. The model included fence treatment, period, fence treatment×period, breed, fence treatment×breed and fence treatment×period. Doeling set was the random effect used to test the effect of fence treatment; period was the repeated measure with the subject of breed within group and set. The final analysis was similar, but only involving data of doeling set 2, which had been previously used in the LS design (abbreviated as Period 1&6-set 2). The model included fence treatment, period, fence treatment×period, breed and fence treatment×breed. Period was a repeated measure with the subject of breed within group.

Normality was evaluated as noted earlier for CRD and LS analyses. Most variables were normally distributed, except for the percentage of doelings exiting in 2 and 6 h for CRD and time of exit in 1 h, number of shocks and the percentage of doelings receiving a second shock for LS. Transformations addressed earlier were employed, but none resulted in normal distribution of these variables and in some cases caused other variables not to be normally distributed. The seven covariance structures stated earlier were evaluated for the percentage of doelings exiting and receiving a first shock in 1 h for CRD and LS, resulting in use of VC. Satterwaite approximate was used to calculate denominator degrees of freedom.

The lsmeans were estimated, and means separation was by Least Significant Difference when overall effects were significant (i.e. protected F-test; p<0.05). Fence treatment main effect means are presented in tables regardless of significance (p<0.05) except when an interaction involving fence treatment was significant, in which case interaction means appear. Other main effect means are listed only when the overall effect was significant, and interaction means are presented when the interaction was significant. Numbers of individual animal observations and p values of variables for which means are presented are given in tabular form. Means for some variables are not given in tables. For example, as noted before, there were no significant effects in shock reaction. Also, low numbers of observations for some variables precluded the specified statistical analyses (e.g. time of shock 2).

Results

Experiment 1

There were no interactions between main effects (p>0.05). The number of shocks was greater (p<0.05) for 4S-LV than for other treatments except 1S-MH-HV, which resulted from a similar percentage of doelings receiving a first shock but the greatest percentage receiving a second shock among treatments for 4S-LV (p < 0.05; Table 1). Period length did not affect the number of shocks or percentage of doelings shocked. Both the number of shocks and percentage of doelings shocked once ranked (p < 0.05) period 1 > 2>3, 4, 5 and 6. The ranking in the percentage of does shocked twice was the same, although the value for period 2 was not different from other periods. Very few doelings received shocks in periods 3–6. Likewise, the percentage of doelings receiving a second shock, even in period 1 when over one-half of doelings received a first shock, was quite low.

Table 1. Effects of length of periods in Latin squares on shock and exit of yearling meat goat doelings in pens with barb wire and different electric fence strand treatments (Experiment 1; lsmeans).

			Variable^1
Fence treatment^2	Period length	Period	Number of shocks	Shock 1 (%)	Shock 2 (%)	Shock 1 (min)	Exit in 2 h (%)	Exit in 6 h (%)
4S-LV			0.26^b	20	9^b	18	8	8
2S-HV			0.15^a	13	1^a	63	2	4
2S-LV			0.14^a	13	0^a	27	2	4
1S-LH-LV			0.16^a	14	2^a	38	2	4
1S-HH-LV			0.17^a	16	3^a	44	4	10
1S-MH-HV			0.21^ab	16	2^a	1	13	17
SE			0.034	2.8	2.0	17.1	3.2	3.4
	7 days		0.18	15	17	33	6	10
	2–3 days		0.18	16	3	31	4	6
	SE		0.025	1.9	0.9	11.8	2.5	2.3
		1	0.63^c	54^c	8^a	17	21^b	27^c
		2	0.29^b	25^b	4^ab	28	8^a	10^b
		3	0.06^a	4^a	2^a	35	2^a	6^ab
		4	0.06^a	6^a	0^a	22	0^a	4^ab
		5	0.00^a	0^a	0^a	–	0^a	0^a
		6	0.04^a	4^a	0^a	36	0^a	0^a
		SE	0.044	4.3	2.0	15.6	3.2	3.4
^1 n=total number of individual animals; 72 for Number of shocks, Shock 1 (%), and Shock 2 (%) in a 2-h continuous observation period, Exit in 2 h (%) and Exit in 6 h (%); n=27 for Shock 1 (min); number of shocks is the average per animal; percentages relate to the total number of doelings.
^24S-LV, four strands of electric fence and low voltage; 2S-HV, two strands and high voltage; 2S-LV, two strands and low voltage; 1S-LH-LV, one strand, low height, and low voltage; 1S-HH-LV, one strand, high height, and low voltage; 1S-MH-HV, one strand, medium height, and high voltage.
^a,b,cMeans in a column within fence treatment, period length or period grouping without a common superscript letter differ (p<0.05).

There were no differences among fence treatments or between period lengths in the percentage of doelings exiting in 2 or 6 h (Table 1). The percentage of doelings that exited pens in 2 and 6 h was greatest among periods (p<0.05) in period 1. No doelings exited pens in 2 or 6 h in periods 5 or 6.

Experiment 2

Exit

Numbers of individual animal observations and p values are presented in Table 2. There were very few exits in the second hour; therefore, exit in the first hour was addressed (Table 3). There were no differences among fence treatments in exit with any of the methods of analysis. Period did not affect exit with the LS method. Nonetheless, pen exit was 29 percentage units lower for the LS vs. CRD approach (p < 0.05), and for the Period 1&6-set 2 analysis exit was 30 percentage units less in period 6 than 1 (p<0.05). For the Period 1&6-set 2 analysis, pen exit was 35 percentage units greater for Spanish vs. Boer (p<0.05); however, a similar breed difference did not occur with the other four analysis methods. There was an interaction (p<0.05) between fence treatment and breed in exit time with CRD and Period 1&3-set 1&3 analyses, although all means were <4 min and most were <1 min

Table 2. p-Values of effects of experimentation method and breed on exit and shock of growing meat goat doelings in pens with barb wire and different electric fence treatments (Experiment 2).

			Source of variation (p value)^d
Variable^a	Analysis^b	n ^c	FT	Prd	FT×Prd	Brd	FT×Brd	Met	Met×FT	Met×Brd
Exit in 1 h (%)	CRD	60	0.4362			0.4226	0.4362
	LS	100	0.0912	0.1740		0.3900	0.3748
	CRD-LS	160	0.0668			0.3304	0.4306	0.0001	0.6129	0.6518
	Period 1&6-set 1&3	120	0.4645	0.0323	0.4645	0.5753	0.7189
	Period 1&6-set 2	40	0.1141	0.0081	0.0722	0.0081	0.0722
Exit in 2 h (%)	CRD	60	0.4362			0.4226	0.4362
	LS	100	0.1266	0.1465		0.3169	0.7124
	CRD-LS	160	0.0962			0.2596	0.7595	0.0005	0.7523	0.6568
	Period 1&6-set 1&3	120	0.4269	0.3273	0.4269	0.3273	0.4269
	Period 1&6-set 2	40	0.1141	0.0081	0.0722	0.0081	0.0722
Exit in 1 h (min)	CRD	57	0.0173			0.2216	0.0044
	LS	66	0.5890	0.1468		0.1541	0.3900
	CRD-LS	123	0.7589			0.1337	0.4933	0.2680	0.4169	0.1686
	Period 1&6-set 1&3	100	0.2053	0.2804	0.3969	0.5096	0.0426
	Period 1&6-set 2^e	28
Number of shocks	CRD	60	0.0012			0.1290	0.0510
	LS	100	0.0001	0.1549		1.0000	0.0876
	CRD-L:S	160	0.0001			0.1342	0.0092	0.3423	0.6333	0.0662
	Period 1&6-set 1&3	120	0.0103	0.0691	0.5416	0.0257	0.0920
	Period 1&6-set 2	40	0.1405	0.0127	0.2955	0.5549	0.6513
Shock 1 (%)	CRD	60	0.0008			0.1705	0.1145
	LS	100	0.0001	0.5410		0.8276	0.0884
	CRD-LS	160	0.0001			0.3616	0.0174	0.6838	0.4880	0.1574
	Period 1&6-set 1&3	120	0.0162	0.3885	0.8463	0.0515	0.1996
	Period 1&6-set 2	40	0.2411	0.0192	0.4090	0.7210	0.7392
Shock 2 (%)	CRD	60	0.0571			0.1835	0.1249
	LS	100	0.1414	0.0083		0.5304	0.7568
	CRD-LS	160	0.0021			0.0814	0.0715	0.0028	0.1316	0.0922
	Period 1&6-set 1&3	120	0.2422	0.0122	0.2422	0.1168	0.3461
	Period 1&6-set 2	40	0.0804	0.0301	0.0804	0.3632	0.4857
Exit at shock (%)	CRD	60	0.0112			0.5286	0.2122
	LS	100	0.0001	0.0671		0.5216	0.1435
	CRD-LS	160	0.0001			0.9497	0.0211	0.8386	0.3659	0.2651
	Period 1&6-set 1&3	120	0.0714	0.3080	0.9145	0.1313	0.4711
	Period 1&6-set 2	40	0.4539	0.0580	0.6892	0.4513	0.4539
Exit, no shock (%)	CRD	60	0.0008			0.1705	0.1145
	LS	100	0.0001	0.3051		0.4050	0.7648
	CRD-LS	160	0.0001			0.1898	0.5295	0.0002	0.6308	0.4004
	Period 1&6-set 1&3	120	0.0075	0.3451	0.9343	0.0660	0.3182
	Period 1&6-set 2	40	0.0537	1.0000	0.0537	0.1019	0.1484
Exit, shock 1 (%)	CRD	60	0.1861			1.0000	0.4468
	LS	100	0.0004	0.02242		0.6387	0.1833
	CRD-LS	160	0.0011			0.7840	0.0371	0.2604	0.2540	0.6581
	Period 1&6-set 1&3	120	0.2927	0.7958	0.9909	0.4401	0.8919
	Period 1&6-set 2	40	0.7739	0.1565	0.7056	0.7524	0.6422
Exit, shock 2 (%)	CRD	60	0.3489			0.4665	0.5427
	LS	100	0.4241	0.4241		0.3739	0.4241
	CRD-LS	160	0.0573			0.5437	0.4213	0.0037	0.1707	0.1506
	Period 1&6-set 1&3	120	0.2422	0.0122	0.2422	0.1168	0.3461
	Period 1&6-set 2	40	0.4857	0.3632	0.4857	0.3632	0.4857
^aPercentages relate to the total number of doelings; Exit in 1 h (min) is the average time for exit during the 1-h continuous observation period.
^bCRD, completely randomised design with the three sets in period 1; LS, Latin square in periods 1–5 with set 2; CRD-LS, comparison of the completely randomised design in period 1 with observations of set 2 in the 5×5 Latin square; Period 1&6-set 1&3, repeated measures analysis with sets 1 and 3 in periods 1 and 6; Period 1&6-set 2, repeated measures analysis with set 2 in periods 1 and 6.
^cTotal number of individual animals.
^dFT, fence treatment (2S-HV = two strands of electric fence and high voltage; 2S-LV = two strands and low voltage; 1S-LH-LV = one strand, low height, and low voltage; 1S-HH-LV = one strand, high height, and low voltage; 1S-MH-HV = one stand, medium height, and high voltage); Prd, period (period 1 = completely randomised design in period 1 with the three sets of 20 doelings; periods 1–5 = 5×5 Latin square with set 2; period 6 = repeat of period 1 following completion of the Latin square with set 2); Set, three sets of 20 doelings (sets 1 and 3 used in periods 1 and 6; set 2 used in all periods); Brd, breed (Boer and Spanish); Met, experimentation method or experimental design (CRD = completely randomised design with the three sets of doelings in period 1; LS = Latin square in periods 1–5 with set 2).
^eAnalysis stopped because of too many likelihood evaluations.

Shocks

No animals incurred three shocks. Electric strand height affected the number of shocks with some analyses, with generally lower values for the fence treatments with one strand at the low or medium height (Table 4). In accordance, the percentage of doelings receiving a first shock with the CRD analysis was lower (p<0.05) for 1S-HH-LV than for 2S-HV, 2S-LV and 1S-LH-LV. Findings for the LS analysis were fairly similar. Although means of first shock for the LS analysis were generally lower than those for CRD, neither LS period nor method with the CRD-LS analysis had a significant effect. With this latter analysis, there was a fence treatment×breed interaction (p<0.05), with lower values for Spanish vs. Boer for most treatments but a much greater value for Boer with the 1S-LH-LV fence treatment. For the Period 1&6-set 2 analysis, the percentage of doelings receiving a first shock was 45 percentage units greater (p < 0.05) in period 1 vs. 6, with no differences among fence treatments. However, period with the Period 1&6-set 1&3 analysis did not have a significant effect.

Table 3. Effects of experimentation method and genotype on exit of growing meat goat doelings from pens with barb wire and different and electric fence strand treatments (Experiment 2; lsmeans).

		Period^1		Breed^2		Fence treatment^3						Method^4
Variable^5	Analysis^6	1–6	Mean	B, S	Mean	2S-HV	2S-LV	1S-LH-LV	1S-HH-LV	1S-MH-HV	SE	CRD	LS	SE
Exit in 1 h (%)	CRD					92	92	92	100	100	6.5
	LS					75	70	40	70	75	13.0
	CRD-LS					91	82	60	90	80	10.2	95^b	66^a	8.7
	Period 1&6-set 1&3	1	100^b			81	88	81	100	100	9.8
		6	80^a
		SE	6.2
	Period 1&6-set 2	1	85^b	B	55^a	50	63	88	75	75	7.9
		6	55^a	S	85^b
		SE	5.0	SE	5.0
Exit in 2 h (%)	CRD					92	92	92	100	100	6.5
	LS					75	70	45	80	80	12.3
	CRD-LS					89	82	63	95	84	9.8	95^b	70^a	7.5
	Period 1&6-set 1&3					100	88	100	100	100	5.6
	Period 1&6-set 2	1	85b	B	55^a	50	63	88	75	75	7.9
		6	55a	S	85^b
		SE	5.0	SE	5.0
Exit in 1 h (min)	CRD			B		0.58	3.25	0.29	0.29	0.25	0.430
				S		0.46	0.25	1.04	0.29	0.25
	LS					0.50	1.39	1.22	3.24	2.23	1.383
	CRD-LS					0.40	1.47	0.61	1.59	1.45	0.881
	Period 1&6-set 1&3			B		0.62	1.63	0.23	0.28	0.25	0.333
				S		0.59	0.25	0.91	0.28	0.34
	Period 1&6-set 2^7
^1Period 1, completely randomised design with three sets of 20 doelings; periods 1–5, Latin square with set 2; period 6, repeat of period 1 following completion of the Latin square with set 2.
Three sets of 20 doelings; sets 1 and 3 used in periods 1 and 6; set 2 used in all periods.
^2B, Boer; S, Spanish.
^32S-HV, two strands of electric fence and high voltage; 2S-LV, two strands and low voltage; 1S-LH-LV, one strand, low height and low voltage; 1S-HH-LV, one stand, high height and low voltage; 1S-MH-HV, one strand, medium height and high voltage.
^4CRD, completely randomised design with the three sets of doelings in period 1; LS, 5×5 Latin square in periods 1–5 with set 2.
^5Percentages relate to the total number of doelings; Exit in 1 h (min) is the average time for exit during the 1-h continuous observation period.
^6CRD, completely randomised design with the three sets of doelings in period 1; LS, Latin square in periods 1–5 with set 2; CRD-LS, comparison of the completely randomised design in period 1 with observations of set 2 in the 5×5 Latin square; Period 1&6-set 1&3, repeated measures analysis with sets 1 and 3 in periods 1 and 6; Period 1&6-set 2, repeated measures analysis with set 2 in periods 1 and 6.
^7Analysis stopped because of too many likelihood evaluations.
^a,b,cMain effect means without a common superscript letter differ (p<0.05).

There were no effects of fence treatment on the percentage of doelings receiving a second shock, other than for the CRD-LS analysis (Table 4). Although the percentage of does receiving a second shock in period 1 was not great, it was higher than in other periods, as indicated by period differences (p<0.05) with LS, Period 1&6-set 1&3, and Period 1&6-set 2 analyses. Likewise, with the CRD-LS analysis the percentage was greater for the CRD approach than for the LS method (p<0.05).

Table 4. Effects of experimentation method and genotype on shocks of growing meat goat doelings in pens with barb wire and different electric fence strand treatments (Experiment 2; lsmeans).

		Period^1		Breed^2		Fence treatment^3						Method^4
Variable^5	Analysis^6	1–6	Mean	B, S	Mean	2S-HV	2S-LV	1S-LH-LV	1S-HH-LV	1S-MH-HV	SE	CRD	LS	SE
Number of shocks	CRD					0.75^b	0.83^b	0.75^b	0.00^a	0.25^a	0.179
	LS					0.80^c	0.75^c	0.45^b	0.05^a	0.15^a	0.103
	CRD-LS			B		0.89	1.09	0.43	0.08	0.28	0.125
				S		0.66	0.49	0.77	0.00	0.12
	Period 1&6-set 1&3			B	0.48^b	0.38^ab	0.63^c	0.50^b	0.00^a	0.25^ab	0.129
				S	0.23^a
				SE	0.092
	Period 1&6-set 2	1	0.65^b			0.75	0.50	0.38	0.00	0.13	0.177
		6	0.06^a
		SE	0.112
Shock 1 (%)	CRD					50^bc	58^c	58^c	0^a	25^ab	14.1
	LS					70^b	70^b	45^b	5^a	15^a	9.3
	CRD-LS			B		68	81	34	6	27	10.4
				S		53	47	70	0	12
	Period 1&6-set 1&3					31^b	50^b	38^b	0^a	25^ab	10.6
	Period 1&6-set 2	1	50^b			50	38	38	0	13	14.8
		6	5^a
		SE	9.4
Shock 2 (%)	CRD					25	25	17	0	0	7.5
	LS	1	15^b			10	5	0	0	0	3.5
		2	0^a
		3	0^a
		4	0^a
		5	0^a
		SE	3.5
	CRD-LS					17^b	15^b	9^ab	0^a	0^a	4.1	13^b	3^a	2.8
	Period 1&6-set 1&3	1	13^b			6	13	13	0	0	5.2
		6	0^a
		SE	3.3
	Period 1&6-set 2	1	15^b			25	13	0	0	0	5.6
		6	0^a
		SE	3.5
^1Period 1, completely randomised design with three sets of 20 doelings; periods 1–5, Latin square with set 2; period 6, repeat of period 1 following completion of the Latin square with set 2.
^2B, Boer; S, Spanish.
^32S-HV, two strands of electric fence and high voltage; 2S-LV, two strands and low voltage; 1S-LH-LV, one strand, low height and low voltage; 1S-HH-LV, one stand, high height and low voltage; 1S-MH-HV, one strand, medium height and high voltage.
^4CRD, completely randomised design with the three sets of doelings in period 1; LS, 5×5 Latin square in periods 1–5 with set 2.
^5Percentages relate to the total number of doelings; Exit in 1 h (min) is the average time for exit during the 1-h continuous observation period.
^6CRD, completely randomised design with the three sets of doelings in period 1; LS, Latin square in periods 1–5 with set 2; CRD-LS, comparison of the completely randomised design in period 1 with observations of set 2 in the 5×5 Latin square; Period 1&6-set 1&3, repeated measures analysis with sets 1 and 3 in periods 1 and 6; Period 1&6-set 2, repeated measures analysis with set 2 in periods 1 and 6.
^a,b,cMain effect means without a common superscript letter differ (p < 0.05).

Exit at shock

Variability in the percentage of doelings that exited in 1 h with and without a shock was slightly greater for the CRD vs. LS analysis (Table 5). Furthermore, differences among fence treatments varied slightly between CRD and LS analyses. There were no differences among fence treatments with either of the Period 1&6 analyses, although means were generally lower for the set 2 than set 1&3 analysis. The percentage of doelings exiting without shock was considerably greater for the CRD vs. LS method (p<0.05). Findings for percentages of doelings exiting upon first and second shocks were generally in accordance with the total number of exits with shock.

Table 5. Effects of experimentation method and genotype on shocks of growing meat goat doelings and exit from pens with barb wire and different electric fence strand treatments (Experiment 2; lsmeans).

		Period^1		Breed^2	Fence treatment^3						Method^4
Variable^5	Analysis^6	1–6	Mean	B, S	2S-HV	2S-LV	1S-LH-LV	1S-HH-LV	1S-MH-HV	SE	CRD	LS	SE
Exit at shock (%)	CRD				42^b	50^b	50^b	0^a	25^ab	13.4
	LS				65^b	55^b	25^a	0^a	15^a	9.9
	CRD-LS			B	57	61	15	0	26	11.7
				S	52	44	59	0	10
	Period 1&6-set 1&3				31	44	38	0	25	10.8
	Period 1&6-set 2				38	25	25	0	13	13.7
Exit, no shock (%)	CRD				50^ab	42^a	42^a	100^c	75^bc	14.1
	LS				10^a	15^a	15^a	70^b	60^b	10.6
	CRD-LS				33^a	29^a	26^a	86^b	65^b	8.7	62^b	34^a	6.0
	Period 1&6-set 1&3				50	44	44	100	75	12.8
	Period 1&6-set 2				13	38	63	75	63	11.2
Exit, shock 1 (%)	CRD				25	33	33	0	25	13.9
	LS				60^b	55^b	25^a	0^a	15^a	10.5
	CRD-LS			B	46	47	7	1	25	10.8
				S	40	41	51	1	13
	Period 1&6-set 1&3				25	31	25	0	25	10.7
	Period 1&6-set 2				25	25	25	0	13	16.8
Exit, shock 2 (%)	CRD				17	17	17	0	0	8.3
	LS				5	0	0	0	0	2.2
	CRD-LS				11	8	8	0	0	3.5
	Period 1&6-set 1&3	1	13^b		6	13	13	0	0	5.2
		6	0^a
		SE	3.3
	Period 1&6-set 2				13	0	0	0	0	5.6
^1Period 1, completely randomised design with three sets of 20 doelings; periods 1–5, Latin square with set 2; period 6, repeat of period 1 following completion of the Latin square with set 2.
^2B, Boer; S, Spanish.
^32S-HV, two strands of electric fence and high voltage; 2S-LV, two strands and low voltage; 1S-LH-LV, one strand, low height and low voltage; 1S-HH-LV, one strand, high height and low voltage; 1S-MH-HV, one strand, medium height and high voltage.
^4CRD, completely randomised design with the three sets of doelings in period 1; LS, 5×5 Latin square in periods 1–5 with set 2.
^5Percentages relate to the total number of doelings.
^6CRD, completely randomised design with the three sets of doelings in period 1; LS, Latin square in periods 1–5 with set 2; CRD-LS, comparison of the completely randomised design in period 1 with observations of set 2 in the 5×5 Latin square; Period 1&6-set 1&3, repeated measures analysis with sets 1 and 3 used in periods 1 and 6; Period 1&6-set 2, repeated measures analysis with set 2 in periods 1 and 6.
^a,b,cMain effect means without a common superscript letter differ (p<0.05).

Discussion

Fence treatment

Based on results of Experiment 1 alone, a fence treatment of barb wire without electric fence strands or with strands at 0 kV would be useful to validate or verify that goats continue to examine fence treatments and have desire to exit evaluation pens. But, the activities undertaken before Experiment 2 to adapt doelings for exit from evaluation pens, as well as results of Experiment 2, do not support merit of including such a treatment.

Even though pen exit was very high in Experiment 2, the low number of shocks for the 1S-HH-LV fence treatment suggests little utility from its inclusion with goats of this size and BW. The lack of differences in Experiment 1 between the 1S-HH-LV and other fence treatments, in contrast to results of Experiment 2, could have been due to the different animal characteristics. But, the low occurrence of shock and pen exit in Experiment 1 for all treatments prevents conclusively attributing these experiment differences to animal type. Relatedly, based on the smaller size of doelings in Experiment 2 than 1, it might be thought that electric fence strands should have been lower in Experiment 2. However, conventional heights based on barb wire strand location were employed. That is, LH was mid-way between the ground and first strand of barb wire, HH was mid-way between the first and second barb wire strands, and MH was just below the second barb wire strand. The generally lower number of shocks and percentage of doelings receiving shock in Experiment 2 for 1S-MH-HV compared with treatments having an electric fence strand at LH implies that either this treatment could be omitted or should be modified, such as decreasing distance from the ground to a height between LH and MH. Placement of an electric fence strand any lower than 15.2 cm would increase problems in fence maintenance via the necessity of vegetation removal to minimise grounding and in installation to avert ground contact with uneven terrain. In contrast, livestock producers have more interest in higher placement of the lowest electric fence strand.

Experimentation method

Low pen exit in Experiment 1, particularly in latter periods of the LS, indicated need for more thorough and repeatable training or adaptation to electric fence and conditions imposed during the experiment. Procedures employed before Experiment 2 presumably did contribute to much different findings than in Experiment 1. Nonetheless, results of Experiment 2 agree with those of Experiment 1 in that LS approaches, as specifically applied in these experiments, may not be appropriate. This is exemplified by period effects in Experiment 1, very low levels of shock and pen exit in Experiment 1, very high exit in Experiment 2 and absence of expected differences among fence treatments in shock and exit during both experiments. The percentage of doelings exiting pens in 1 h with the LS approach in Experiment 2, though lower than with the CRD approach, was still relatively high and did not differ among fence treatments as expected. Moreover, levels of pen exit in both periods 1 and 6 of Experiment 2 were very high. Nonetheless, marked differences in results from LS of Experiments 1 and 2 do not conclusively indicate that a LS approach cannot be used. But, development of an acceptable method would require further change in one or most likely a number of conditions.

Experimental conditions

One or more of the conditions of Experiment 1 modified for Experiment 2 could have been responsible for the remarkable shift from very low to high pen exit regardless of fence treatment. It is also noteworthy that in neither experiment, with such different levels of pen exit, were differences among fence treatments as expected, except perhaps for the number of shocks for treatments with different electric strand height in Experiment 2. Therefore, in addition to alterations of procedures to improve efficacy of a LS or CRD approach, simultaneous attention should be given to changes for achieving expected differences among fence treatments.

Most notable conditions altered between Experiments 1 and 2 were in (1) prior exposure to electric fence before adaptation; (2) repeatable training to electric fence before the experiment; (3) training to promote subsequent pen exit by placement in pens with barb wire alone or plus one strand of electric fence at 0 kV; (4) removal of the centre steel post to allow greater wire movement upon animal contact; (5) length of fasting; (6) use of younger, smaller and lighter goats that should lessen physical impediments to exit by movement under or between fence strands and (7) inclusion of a genotype thought to have a relatively high propensity for pen exit. Further experimentation will be required to partition effects of the various conditions altered. There are other factors to be considered as well. For example, although soil moisture level was not measured, ground in the study area was noticeably drier in Experiment 2 vs. 1. Wetting of the ground near and under electric fence strands in Experiment 2 could have been inadequate and contributed to the high percentage of doelings exiting rapidly without shock. Subsequent research should entail a constant and higher soil moisture level than in Experiment 2, also perhaps including the entire evaluation pen and surrounding area to ensure ample grounding upon contact with electric fence strands.

One of the suggestions arising from results of Experiment 1 was that a period length longer than 1 wk might be required for memory of exposure to electric fence to dissipate to an extent allowing subsequent evaluation with minimal or no carryover effects. However, before Experiment 2 was conducted, the same doelings used in Experiment 1 were employed in the first three periods of two 6×6 LS comparing 1- and 2-wk lengths. This study was terminated because of little or no pen exit regardless of period length, similar to findings in the latter periods of Experiment 1. Hence, for Experiment 2 it was assumed that there would be no appreciable benefit from a period length longer than 1 wk. Lower exit in 1 h during week 6 vs. 1 for the Period 1&6-set 1&3 analysis of Experiment 2 is supportive and indicates influence even after 5 wk, although this was the only variable with such a difference for this method of analysis. Also, it is not readily apparent if this difference between two very high levels of pen exit would be applicable to conditions that would result in some treatments having much lower levels of exit.

In contrast to anecdotal evidence leading to inclusion of Spanish doelings, it would not appear that this genotype has a markedly greater propensity for pen exit than the Boer genotype. But, it is unclear if findings would be similar with change in some of the conditions noted above and below.

Another issue in need of attention, previously addressed only in regards to period length, is management of animals between measurement periods. For example, continuing to expose goats to electric fence with a practical voltage level between times of measurement might decrease immediate pen exit regardless of fence treatment. In fact, many goat producers comment that after animals are thoroughly accustomed to electric fence strands in pasture areas, exit does not occur or does so only after a long period of time subsequent to undetected grounding or power disconnect with 0 kV. Therefore, future experimentation should address impact of continued exposure to electric fence strands with typical voltage between periods of testing. Moreover, some experiments with multiple periods include ‘washout’ periods, the intent of which is for effects of treatments imposed in the preceding period to dissipate to or near zero for prevention of carryover effects. Situation in ‘home’ pastures without electric fence strands at typical voltage in the present experiments may not have been an appropriate washout strategy. Alternative washout activities include exposure to a pen with barb wire only or plus 1 strand of electric fence at 0 kV as in the preliminary phase of Experiment 2. However, based on results of Experiment 2, this would not seem a logical choice. Rather, exposure to an electric fence treatment that should be highly prohibitive to pen exit, such as four strands at high, moderate or low voltage, could be a consideration for washout as well as training before the experiment. Such a practice was not considered for Experiment 2 because results of Experiment 1 suggested that changes be made to increase pen exit, albeit not to the extent realised.

Summary and conclusions

Expected differences among fence treatments in pen exit were not observed in either experiment. Percentages of doelings receiving shocks and exiting pens were low with yearling meat goat doelings in a LS, although values were greater in early than later periods. Presumably because of a number of changes in conditions, pen exit was very high in the second experiment with growing doelings used in both CRD and LS approaches. Effects of repeated exposure to treatments were again evident with doelings used in a LS and in one variable, pen exit during 1 h of observation, for other doelings with 5 wk between two measurement periods. Further changes in conditions, such as in adaptation before experiments and exposure to electric fence between measurement periods, may be necessary to develop a method of evaluating efficacy of different additions of electric fence strands to cattle barb wire fence for goat containment.

Acknowledgements

This experiment was supported by USDA Project Number 2010-38814-21561.