1. Introduction 1. 引言

Riparian zones are critical to ecosystems because they can buffer and filter surface water flows, and capture sediment (Government of Western Australia 2000). Meanwhile, the riparian zones are also critical for the stability of river banks and ecological habitats, environmental aesthetics as well as species diversity (Queensland Government 2013). There is a view that in the long run, the destruction of the riparian zone by overgrazing cannot be ignored (Davis, 1982).
河岸带对生态系统至关重要，因为它们可以缓冲和过滤地表水流，并捕获沉积物（西澳大利亚政府 2000 年）。同时，河岸带对于河岸和生态栖息地的稳定性、环境美学以及物种多样性也至关重要（昆士兰政府 2013 年）。有一种观点认为，从长远来看，过度放牧对河岸区的破坏不容忽视（Davis， 1982）。

From the team planning report, it can be concluded that the diversity, quantity, and coverage of vegetation are directly affected by trampling and grazing (Dorrough et al. 2004; Roath & Kruege 1982). Additionally, trampling and grazing also increase pH , increase soil moisture, and increase soil porosity (Whalen et al. 2000; Mullins&Fraser, 1980; Trimble & Mendel, 1995). Our original hypothesis was, “The presence of livestock in riparian zones has a negative impact on natural vegetation species.” During the field trip, we only identified nonnative species. The prevalence of non-native species might be due to our choice of sites. In fact, all of our sites are close to the habitats or areas that humans have intervened in the past. Therefore, we changed a new hypothesis, which is the presence of livestock in riparian zones has a negative impact on soil quality and vegetation quality.
从团队规划报告中可以得出结论，植被的多样性、数量和覆盖率直接受到践踏和放牧的影响（Dorrough 等人，2004 年;Roath & Kruege 1982）。此外，践踏和放牧也会增加 pH 值，增加土壤水分，并增加土壤孔隙度（Whalen 等人，2000 年;Mullins&Fraser，1980 年;Trimble & Mendel，1995 年）。我们最初的假设是，“河岸区牲畜的存在对自然植被物种有负面影响。在实地考察期间，我们只发现了非本地物种。非本地物种的普遍存在可能是由于我们选择的地点。事实上，我们所有的工厂都靠近人类过去干预过的栖息地或区域。因此，我们改变了一个新的假设，即河岸区牲畜的存在对土壤质量和植被质量有负面影响。

We chose to quantitatively evaluate the impact of the introduction of livestock on soil quality and vegetation quality based on the Control-Impact monitoring design. According to the team camps report, we found that the results of the vegetation diversity were opposite to our conceptual model, thus changing the alternative hypothesis. This report will focus on site selection, data collection, as well as data analysis, which including hypothesis tests and power analysis. Moreover, it uses the works of literature to explain the results and point out how to improve in the discussion section.
我们选择根据控制影响监测设计定量评估引入牲畜对土壤质量和植被质量的影响。根据团队营地报告，我们发现植被多样性的结果与我们的概念模型相反，从而改变了替代假设。本报告将侧重于选址、数据收集以及数据分析，包括假设检验和功效分析。此外，它使用文学作品来解释结果并在讨论部分指出如何改进。

2. Method 2. 方法

2.1 Field sites 2.1 野外地点

Due to the limited time of the Dookie Filedtrip, we cannot use the before-after control-impact (BACI) method. We used spatial changes instead of temporal changes and used replication to avoid spurious results. Therefore, this project selects two impact and two control sites. (More detailed information is available in the Team Camps Report.)
由于 Dookie Filedtrip 的时间有限，我们不能使用前后控制影响（BACI）方法。我们使用了空间变化而不是时间变化，并使用复制来避免虚假结果。因此，本项目选择了两个影响站点和两个控制站点。（更多详细信息可在团队训练营报告中找到。

Table 1. Locations and pictures of sites.
表 1.地点的位置和图片。

	Impact 冲击	Control 控制
Site 1 站点 1	Broken River K1 Streamside Reserve ( $36^{\circ} 36^{\circ} 09^{'' S} 145^{\circ} 58^{' 2} 21^{'' E}$ ) 布罗肯河 K1 河畔保护区（ $36^{\circ} 36^{\circ} 09^{'' S} 145^{\circ} 58^{' 2} 21^{'' E}$ ）	Broken River K25 Streamside Reserve (36³0'33.9"S 14557'20.3"E) 布罗肯河 K25 河边保护区（36³0'33.9“S 14557'20.3”E）
Site 2 站点 2	Nearby the Goomalibee Bridge, on the south side of the broken river ( $36^{\circ} 27^{'} 40^{'' S} 145^{\circ} 51^{'} 39^{'' E}$ ) 附近 Goomalibee 大桥，在断河南侧（ $36^{\circ} 27^{'} 40^{'' S} 145^{\circ} 51^{'} 39^{'' E}$ ）	Nearby the Goomalibee Bridge, on the south side of the broken river ( $36^{\circ} 27^{'} 39^{'' S} 145^{\circ} 51^{'} 41^{''} E)$ 在 Goomalibee 大桥附近，在断河南侧（ $36^{\circ} 27^{'} 39^{'' S} 145^{\circ} 51^{'} 41^{''} E)$

2.2 Data collection 2.2 数据收集

To satisfy the replication and randomisation, we randomly chose three quadrats at each site to collect data on vegetation and soil. We use field forms and lab forms to record data and upload them into Google drive to share with the group. (Monitoring design, sampling method, and more details are shown in the Team Camps Report)
为了满足复制和随机化，我们在每个地点随机选择了三个样方来收集植被和土壤的数据。我们使用现场表单和实验室表单来记录数据并将其上传到 Google Drive 以与小组共享。（监测设计、抽样方法和更多详细信息显示在团队训练营报告中）

Table 2. Sample collection.
表 2.样本采集。

测量端点

Measurement

Endpoint

Sampling type 采样类型

Numbers of Sampls Sampls 的数量

Impact site 影响部位

Control site 控制部位

total 总

Soil pH 土壤 pH 值

Field work 实地考察

Soil moisture 土壤水分

田野工作（收集数据）和实验室工作

Field work (collect

data) & Lab work

Soil porosity 土壤孔隙度

田野工作（收集数据）和实验室工作

Field work (collect

data) & Lab work

植被多样性（物种数量）

Vegetation diversity

(Numbers of

species)

Field work 实地考察

Numbers of plants 植物数量

Field work 实地考察

vegetation coverage 植被覆盖率

Field work 实地考察

2.3 Data analysis 2.3 数据分析

2.3.1 Data Integration 2.3.1 数据集成

Check the data quality. Since all the vegetation coverage is calculated by the swap group, we found many results are problematic. Therefore, the results are re-calculated (the results are shown in the Appendix).
检查数据质量。由于所有植被覆盖率都是由交换组计算的，因此我们发现许多结果都是有问题的。因此，结果被重新计算（结果显示在附录中）。
Use Microsoft Excel to merge all subsamples into average data for each site.
使用 Microsoft Excel 将所有子样本合并为每个站点的平均数据。
Use Microsoft Excel to calculate, ‘Standard deviation’, ‘Difference in mean’ and ‘Pooled standard error’.
使用 Microsoft Excel 计算 '标准差'， '均值差异' 和 '合并标准误差'。
Use Microsoft Excel to draw a comparison histogram with average data and error bars is for each measurement endpoint.
使用 Microsoft Excel 绘制包含平均数据的比较直方图，误差线适用于每个测量端点。

2.3.2 Hypothesis Test 2.3.2 假设检验

Since our group’s data is a set of two independent samples that follow a normal distribution, we chose to use T-TEST to compare if there are significant differences between the two samples, which is the most widely used method of statistics (Xu et al., 2017).
由于我们小组的数据是一组两个遵循正态分布的独立样本，我们选择使用 T-TEST 来比较两个样本之间是否存在显著差异，这是使用最广泛的统计方法（Xu et al.， 2017）。

Steps for T-Test: T 检验的步骤：

Specify the null and alternative hypothesis.
指定 null 假设和备择假设。

Table 3. Hypotheses. 表 3.假设。

Assessment Endpoint 评估终点	Measurement Endpoint 测量端点	Null Hypothesis HO 原假设 HO	Alternative Hypothesis HA 备择假设 HA
Soil quality 土壤质量	Soil pH 土壤 pH 值	Impact and Control sites have the same soil pH value. (Or Impact<Control) 影响地点和对照地点具有相同的土壤 pH 值。（或 Impact<Control）	Impact sites have higher soi pH value. 影响部位的 SOI pH 值较高。
	Soil moisture 土壤水分	Impact and Control sites have the same soil moisture value. (Or Impact<Control) 影响地点和控制地点具有相同的土壤含水量。（或 Impact<Control）	Impact sites have higher soi moisture value. 影响部位具有较高的 soi 水分值。
	Soil porosity 土壤孔隙度	Impact and Control sites have the same soil porosity value. (Or Impact>Control) 影响地点和控制地点具有相同的土壤孔隙度值。（或 Impact>Control）	Impact sites have lower soil porosity value. 影响地点的土壤孔隙度值较低。
Vegetation quality 植被质量	Vegetation diversity (Numbers of species) 植被多样性（物种数量）	Impact and Control sites have the same numbers of species. (Or Impact<Control) 影响站点和控制站点具有相同的物种数量。（或 Impact<Control）	Impact sites have higher numbers of species. 撞击地点的物种数量较多。
	Numbers of plants 植物数量	Impact and Control sites have the same numbers of plants. (Or Impact>Control) 影响地点和控制地点的植物数量相同。（或 Impact>Control）	Impact sites have lower numbers of plants. 影响地点的植物数量较少。
	vegetation coverage 植被覆盖率	Impact and Control sites have the same vegetation coverage. (Or Impact>Control) 影响地点和控制地点具有相同的植被覆盖范围。（或 Impact>Control）	Impact sites have lower vegetation coverage. 影响地点的植被覆盖率较低。

Choose significant level $α (0.05)$
选择显著水平 $α (0.05)$

Due to the close relationship between the probability of Type I error (

α

) and Type II error (

β

), choosing an appropriate

α

level is crucial for both t-test and subsequent power analysis. In environmental impact studies, Type II error is more concerned
由于 I 型误差（

α

）和 II 型误差（）的概率之间存在密切关系

β

，因此选择合适的

α

水平对于 t 检验和后续功效分析都至关重要。在环境影响研究中，II 类误差更受关注
based on the precautionary principle. In such circumstances,

β

should be reduced to avoid Type II error. However, for a study with a given sample size and effect size, decreasing

β

will result in an increased

α

(Mudge et al. 2012). Therefore, to reduce the possibility of missing critical environmental impacts, meanwhile, make sure the probability of Type I error occurring within an acceptable range, the significant level

α

is set to 0.05 .
基于预防原则。在这种情况下，

β

应减少以避免 II 型误差。然而，对于具有给定样本量和效应量的研究，减少

β

将导致增加

α

（Mudge 等人，2012 年）。因此，为了减少遗漏关键环境影响的可能性，同时确保 I 类错误发生的概率在可接受的范围内，显着水平

α

设置为 0.05 。
3. Use the tool “data analysis” in Microsoft Excel, to obtain the degree of freedom (df) test statistic test statistic and the p value.
3. 使用 Microsoft Excel 中的“数据分析”工具，获取自由度（df）检验统计量检验统计量和 p 值。
4. Compare p-value to

α (0.05)

to reject or retain the null hypothesis

(H_{0})

.
4. 比较 p 值

α (0.05)

以否定或保留原假设

(H_{0})

。

If p-value

< α

, the null hypothesis

(H_{0})

will be rejected and accept the alternative hypothesis

(H_{A})

.
如果 p-value

< α

，则原假设

(H_{0})

将被拒绝并接受备择假设

(H_{A})

。

If p-value

> α

, the null hypothesis

(H_{0})

will be retained and reject the alternative hypothesis

(H_{A})

.
如果 p-value

> α

，

(H_{0})

将保留原假设并拒绝备择假设

(H_{A})

。

2.3.3 Power Analysis 2.3.3 功率分析

For variables that are not significantly different, power analysis is adopted to assess the probability of committing a Type II error (the probability of accepting null hypothesis if it is false).In addition, the post-hoc power analysis can also help interpret statistical results and identify the power of the

t

-test to detect as significant the environmentally significant effect size (Onwuegbuzie, 2004).
对于差异不显著的变量，采用功效分析来评估犯 II 类错误的概率（如果原假设为假，则接受原假设的概率）。此外，事后功效分析还可以帮助解释统计结果并确定

t

-test 检测环境显著效应大小的能力（Onwuegbuzie，2004）。

Steps for Power Analysis:
幂分析的步骤：

Defining the environmental significant effect size through research.
通过研究确定环境显着效应大小。
Add the environmentally significant effect size to the sample mean of the control sites to get the sample mean at the impact site.
将环境显著性效应大小与对照位点的样本均值相加，得到影响位点的样本均值。
Use G Power, input ‘data variability’, 'significant level $α$ ', 'effect size coefficient $d$ ', and 'sample size $n$ ', then it can obtain power and $β$ .
使用 G 幂，输入 'data variability'、'significant level $α$ '、'effect size coefficient $d$ ' 和 'sample size $n$ '，即可获得幂和 $β$ 。

点6

3. Results 3. 结果

3.1 Hypothesis test 3.1 假设检验

Figure 1. Soil pH. 图 1.土壤 pH 值。
ControlImpact

Figure 2. Soil moisture. 图 2.土壤水分。

Figure 3. Soil porosity. 图 3.土壤孔隙度。

The soil pH slightly increased, and the soil moisture has significantly increased in which the control group is

0.095 (9.5 %)

, and the impact group is

0.166 (16.6 %)

(Figure

1 & 2

). Additionally, the control group has a significant standard-deviation, indicating that the data within the group is too different. Figure 3 shows a decrease in the soil porosity of the impact group. The preliminary results presented in these three figures are the same as our hypothesis. But by numerical value alone, the result is fragile. Therefore, according to our conceptual
土壤 pH 值略有增加，土壤水分显著增加，其中对照组为

0.095 (9.5 %)

，影响组为

0.166 (16.6 %)

（图

1 & 2

）。此外，对照组具有显著的标准差，表明组内的数据差异太大。图 3 显示了冲击组土壤孔隙度的降低。这三个图中呈现的初步结果与我们的假设相同。但仅从数值来看，结果是脆弱的。因此，根据我们的概念
model, a one-tailed t-test is used to observe further whether there are significances between the data.
model 中，使用单尾 t 检验进一步观察数据之间是否存在显著性。

Table 4. T-Test results (soil quality.)
表 4.T 检验结果（土壤质量。

Soil pH 土壤 pH 值

T-Test (two-tailed,

a = 0.05

)
T 检验（双尾，

a = 0.05

）

p -value p 值

results 结果

0.4790

0.2196

Retained

H_{O}

, Rejected

H_{A}

保留

H_{O}

，已拒绝

H_{A}

Power Analysis 功率分析

effect size coefficient d
效应大小系数 d

sample size 样本量

Power 权力

β

0.8538

Impact 2 控件 2

Impact 2

Control 2

0.1493

0.8507

Soil moisture 土壤水分

T-Test (one-tailed,

a = 0.05

)
T 检验（单尾，

a = 0.05

）

t

p -value p 值

results 结果

1.2466

0.1585

Retained

H_{O}

, Rejected

H_{A}

保留

H_{O}

，已拒绝

H_{A}

Power Analysis 功率分析

effect size coefficient d
效应大小系数 d

sample size 样本量

Power 权力

β

1.2452

Impact 2 控件 2

Impact 2

Control 2

0.2187

0.7813

Soil porosity 土壤孔隙度

T-Test (one-tailed,

a = 0.05

)
T 检验（单尾，

a = 0.05

）

t

p -value p 值

results 结果

1.5932

0.0429

Rejected

H_{0}

, Accept

H_{A}

已拒绝

H_{0}

，接受

H_{A}

For both soil pH and soil moisture, their p -value is more than 0.05 ( p -value: soil

pH = 0.2196

, soil moisture

= 0.1585

), which means it retained H 0 and rejected HA (Table 4). In other words, the presence of livestock in riparian zones will not have significant impacts on the soil pH and soil moisture. After the power analysis, the results show that both of the two variables have very low power and high probability of Type II error, which is 0.8507 and 0.7813 (Table 4).
对于土壤 pH 值和土壤湿度，它们的 p 值都大于 0.05（p 值：土壤

pH = 0.2196

、土壤水分

= 0.1585

），这意味着它保留了 H 0 并拒绝了 HA（表 4）。换句话说，河岸区的牲畜不会对土壤 pH 值和土壤湿度产生重大影响。经过功效分析，结果表明，这两个变量的功效都非常低，并且 II 型误差的概率很高，分别为 0.8507 和 0.7813（表 4）。

For soil porosity, the

t

-ratio is 1.5932 , and the p -value is 0.0429 (Table 4). Because the pvalue is less than 0.05 , it rejected H0 but accept HA. Therefore, the presence of livestock in riparian zones will significantly reduce the soil porosity.
对于土壤孔隙度，

t

-ratio 为 1.5932，p -值为 0.0429（表 4）。由于 pvalue 小于 0.05 ，因此它拒绝了 H0 但接受 HA。因此，河岸区牲畜的存在将显着降低土壤孔隙率。

3.1.2 Vegetation quality 3.1.2 植被质量

Figure 4. Vegetation diversity.
图 4.植被多样性。

Figure 6. Vegetation coverage.
图 6.植被覆盖率。

Figure 5. Number of plants.
图 5.植物数量。

From figure 4, it can be seen that the impact group(4) had higher vegetation diversity than the control group (2.5). Figure 5 showed the number of plants between the control group and impact group has a huge difference, which is 202.5 and 72.5 . Finally, the result from vegetation coverage does not show a big difference (Figure 6), but it still the same as our conjecture. Although the Preliminary Results support our hypothesis, we still need to use a one-tailed t-test to validate these statements.
从图 4 中可以看出，影响组（4）的植被多样性高于对照组（2.5）。图 5 显示对照组和影响组之间的植物数量存在巨大差异，分别为 202.5 和 72.5。最后，植被覆盖的结果没有显示出很大的差异（图 6），但它仍然与我们的猜想相同。尽管初步结果支持我们的假设，但我们仍然需要使用单尾 t 检验来验证这些陈述。

Table 5. T-Test results (vegetation quality.)
表 5.T 检验结果（植被质量。

Vegetation diversity (number of species)
植被多样性（物种数量）

T-Test (one-tailed,

a = 0.05

)
T 检验（单尾，

a = 0.05

）

p -value p 值

results 结果

2.17

0.025

Rejected

H_{0}

, Accept

H_{A}

已拒绝

H_{0}

，接受

H_{A}

Number of plants 工厂数量

T-Test (one-tailed,

a = 0.05

)
T 检验（单尾，

a = 0.05

）

p -value p 值

results 结果

1.4008

0.1963

Retained

H_{0}

, Rejected

H_{A}

保留

H_{0}

，已拒绝

H_{A}

Power Analysis 功率分析

effect size coefficient d
效应大小系数 d

sample size 样本量

Power 权力

β

1.4008

2 Impact 2 控制

2 Impact

2 Control

0.2499

0.7501

Vegetation coverage 植被覆盖率

T-Test (one-tailed,

a = 0.05

)
T 检验（单尾，

a = 0.05

）

t

p -value p 值

results 结果

0.1918

0.439

Retained

H_{0}

, Rejected

H_{A}

保留

H_{0}

，已拒绝

H_{A}

Power Analysis 功率分析

effect size coefficient d
效应大小系数 d

sample size 样本量

Power 权力

β

0.2052

2 Impact 2 控制

2 Impact

2 Control

0.0674

0.9326

According to the table 5 , vegetation diversity’s t-ratio is 2.17 , and the p -value is 0.025 , which is less than the significance level

(α = 0.05)

and accepts HA. In other words, the presence of livestock in riparian zones will increase vegetation diversity.
根据表 5 ，植被多样性的 t 比值为 2.17 ，p 值为 0.025 ，小于显著性水平

(α = 0.05)

并接受 HA。换句话说，河岸区牲畜的存在将增加植被的多样性。

The p-value of the number of plants and the vegetation coverage are 0.1963 and 0.439 , both of them are more than 0.05 (Table 5). They Retained the H0 and Rejected HA, which means the presence of livestock in riparian zones will not have significant impacts on the number of plants and the vegetation coverage. Therefore, we use power analysis to calculate the probability of Type II error, table 5 Show that both of them have low power and high

β

(The number of plants: power

= 0.2499, β = 0.7501

; The vegetation coverage: power

= 0.0674

β = 0.9326)

.
植物数量和植被覆盖度的 p 值为 0.1963 和 0.439，均大于 0.05（表 5）。他们保留了 H0 和拒绝了 HA，这意味着河岸区的牲畜不会对植物数量和植被覆盖率产生重大影响。因此，我们使用幂分析来计算 II 型误差的概率，表 5 显示它们都具有低功率和高

β

（植物数量：功率

= 0.2499, β = 0.7501

;植被覆盖率：功率

= 0.0674

，

β = 0.9326)

3.2 Power analysis 3.2 功率分析

Table 6. Power analysis results.
表 6.幂分析结果。

Power Analysis 功率分析

环境 al 显著效应大小

environment

al significant

effect size

效应大小系数 T D

effect

size

coefficien

t d

样本量

sample

size

Power 权力

β

高功率/低

High

Power/

Low

Soil pH 土壤 pH 值

10 %

10.2645

0.5

Impact 2 控件 2

Impact 2

Control 2

0.9995

0.0005

High 高

土壤水分

Soil

moisture

30 %

0.5002

0.5

Impact 2 控件 2

Impact 2

Control 2

0.0994

0.9006

Low 低

土壤孔隙度

Soil

porosity

25 %

4.5637

0.5

Impact 2 控件 2

Impact 2

Control 2

0.8756

0.1244

High 高

Vegetatio n 多样性（

Vegetatio

n diversity

(number

20 %

2.8935

0.05

Impact 2 控件 2

Impact 2

Control 2

0.5937

0.4063

Low 低

species) 物种）

工厂数量

Number

of plants

50 %

2.1719

0.05

Impact 2 控件 2

Impact 2

Control 2

0.4247

0.5753

Low 低

Vegetatio n

Vegetatio

50 %

1.6079

0.05

Impact 2 控件 2

Impact 2

Control 2

0.2941

0.7059

Low 低

coverage 覆盖

Based on literature research, we have given different environmentally significant effect sizes for different measurement endpoints. (It will be discussed in detail in the next section). From the above table, only the power of soil pH and soil porosity is high, which are 0.9995 and 0.8756 . The power of soil moisture, vegetation diversity, number of plants, and vegetation are low, especially the soil moisture, in which the power is only 0.0994 . This result shows that, besides the soil pH and the soil porosity, none of our other inspection designs have powerful features to detect observed effects.
根据文献研究，我们为不同的测量终点给出了不同的环境显着效应大小。（将在下一节中详细讨论）。从上表可以看出，只有土壤 pH 值和土壤孔隙度的功效较高，分别为 0.9995 和 0.8756 。土壤水分、植被多样性、植物数量和植被的功效较低，尤其是土壤水分，其中功效仅为 0.0994。这一结果表明，除了土壤 pH 值和土壤孔隙度之外，我们的其他检测设计都没有强大的功能来检测观察到的效果。

In addition, for two measurement endpoints that we have already found a significant effect in the t-test, we will use post-hoc power analysis to identify the power of the t-test. The power of the soil porosity is 0.8756 , which is less likely to indicate a type II error, which means that
此外，对于我们已经在 t 检验中发现显著效应的两个测量终点，我们将使用事后功效分析来确定 t 检验的功效。土壤孔隙度的幂为 0.8756 ，这不太可能表示 II 型误差，这意味着
the correct conclusion is more likely to be obtained in the t-test. However, vegetation diversity has low power, which s only 0.5937 and has a large possibility of type II errors, so the results obtained in the t-test may not prove it has a significant difference.
在 t 检验中更有可能获得正确的结论。然而，植被多样性的功率较低，仅为 0.5937，且存在 II 型误差的可能性很大，因此 t 检验得到的结果可能无法证明其存在显著差异。

1. Introduction 1. 引言

2. Method 2. 方法

2.1 Field sites 2.1 野外地点

2.2 Data collection 2.2 数据收集

2.3 Data analysis 2.3 数据分析

2.3.1 Data Integration 2.3.1 数据集成

2.3.2 Hypothesis Test 2.3.2 假设检验

Steps for T-Test: T 检验的步骤：

2.3.3 Power Analysis 2.3.3 功率分析

Steps for Power Analysis:幂分析的步骤：

3. Results 3. 结果

3.1 Hypothesis test 3.1 假设检验

3.1.2 Vegetation quality 3.1.2 植被质量

3.2 Power analysis 3.2 功率分析

Steps for Power Analysis:
幂分析的步骤：