Characterizing Temporal Ecophysiology for Chemical Management of Huisache (Acacia Farnesiana [L.] Willd.)

Characterizing Temporal Ecophysiology for Chemical Management of Huisache (Acacia farnesiana [L.] Willd.)

by Pablo C. Teveni III, B.S.A., M.S. A Dissertation In Wildlife, Aquatic, and Wildlands Science and Management

Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY

Robert D. Cox, Ph.D. Chair of Committee Ron Sosebee, Ph.D. Carlos Villalobos, Ph.D. Glen Ritchie, Ph.D. Eric Walden, Ph.D.

Mark Sheridan, Ph.D. Dean of the Graduate School August, 2017

Texas Tech University, Pablo Teveni, August 2017

Acknowledgements

I would like to express my sincere gratitude to those who have helped me with my doctoral dissertation and studies. I thank my major professor Dr. Robert D. Cox, who was always helpful, patient, and optimistic, making even the hardest tasks seem surmountable. I would like to thank the other members of my graduate committee, Dr.

Ron Sosebee, Dr. Carlos Villalobos, Dr. Glen Ritchie, and Dr. Eric Walden, for guidance, assistance and, encouragement. I thank Dr. Terry Blankenship, Dr. Selma Glasscock, and the Welder Wildlife Foundation for sponsoring my studies and giving me the opportunity to be a Welder fellow. I also thank the Texas Coalition Grazing Lands Conservation

Initiative, Dow AgroSciences, DuPont Crop Protection, King Ranch, Martin O’Connor

Cattle Company, and Diebel Cattle Company for their financial and/or material assistance for this project. I would specifically like to thank Clifford Carter, Stephen Diebel, Verl

Cash, Wayne Hanselka, and Stephen Deiss for their help with this project. Finally I would like to thank my sisters Michelle Hernandez, Natalie McDonald, and Laura Teveni my brother Michael Teveni, my mother Sylvia Luna, my father Pablo Teveni, Jr., my aunt Marcelina Teveni, my uncle Robert Teveni, and my fiancé, Andrea De Leon, for their support and encouragement over the course of my dissertation research.

iii

Texas Tech University, Pablo Teveni, August 2017

Table of Contents

Abstract…………………………………………………………………………………...vi

List of Tables.…………………………………………………………………...……...... ix

List of Figures…………………………………………………………………………….xi

Chapter 1. INTRODUCTION………………..…………………………………………...1

Huisache History in South Texas………………………………………..……...... 2

Effect of Huisache on Rangelands………………………………………………...6

Huisache Control Methods………………………………………………………12

Total Nonstructural Carbohydrates and Herbicide Susceptibility……...……….18

Huisache Seed Germination………………………...…………………………...20

Huisache Seedling Resprouting and Cotyledon Latent Meristem

Positioning…………………………………….…………………..……………..21

Objectives………………………………………………………………………..22

Chapter 2. MATERIALS AND METHODS……………………………………………23

Study 1: Effective Herbicide Control of Huisache……………………………...23

Description of Study Sites……………………………………………….23

Phenology……………………………………………………………...... 28

Weather Stations…………………………………………………………28

Carbohydrate Analysis…………………………………………………..29

Chemical Control Demonstration……………………………………...... 30

Analysis ……………...... 31

Study 2: Huisache Seed Germination………………………………………...... 33

Study 3: Huisache Seedling Resprouting and Cotyledon Latent

Texas Tech University, Pablo Teveni, August 2017

Meristem Positioning.…………………………………………………………....34

Analysis ……………………………………………....………………….35

Chapter 3. RESULTS……………………………...………………………….……...... 36

Study 1: Effective herbicide control of huisache…………………….………….36

Carbohydrate Analysis and Chemical Control

Demonstration –ANOVA...... 36

Carbohydrate Analysis and Chemical Control

Demonstration – AIC…………………………………………………….47

Study 2: Huisache Seed Germination…………………………………………………58

Study 3: Huisache Seedling Resprouting and Cotyledon Latent Meristem

Positioning...……………………………………………………………………..60

Chapter 4. Discussion…………………………………………...... 68

Study 1: Effective herbicide control of huisache……………………………………68

Study 2: Huisache Seed Germination…………………………………………...73

Study 3: Huisache Seedling Resprouting and Cotyledon Latent Meristem

Positioning…………………………………………………………………...…..74

Conclusions………………………………………………………..……...……...75

Literature Cited………………………………………………………………..…...... 79

Appendix………………………………………………………………………..……...105

Texas Tech University, Pablo Teveni, August 2017

Abstract Huisache (Acacia farnesiana [L.] Willd.), a pest tree species on rangelands throughout south Texas, is expanding in range and density and displacing more desirable forage grasses and forbs. It resists control via herbicide, fire, or mechanical methods through vigorous resprouting and prolific seed production. The objectives of my research were to 1) characterize the optimal timing and environmental conditions for effective herbicide control of huisache, 2) determine the optimum diurnally-fluctuating temperature conditions for huisache seed germination, and 3) determine the age at which seedlings’ cotyledon latent meristems are buried below the soil surface and plants attain resprouting ability. These objectives were addressed in three experiments. For the first, four huisache-invaded pastures with different soil textures from four counties along the Texas Coastal Bend region were used. On-site weather station data were gathered, and the average phenological stage of huisache was determined each month for each pasture; shrubs from the average phenological stage were used for TNC (total nonstructural carbohydrate) analysis and herbicide treatment. Each month between April 2012 and November 2014, five shrubs per site were excavated, and the root crowns were collected, frozen, dried, ground, and analyzed for TNC. In addition, two herbicide formulations were foliar-applied to five to ten shrubs every month at the four sites from July 2012 to November 2014, and mortality was evaluated following two growing seasons. For the seed germination study, five 20-seed petri dish replicates were used in each of four seed temperature treatments, including: 35/25°C, 30/20°C, 25/15°C, and 20/10°C, under alternating 12-hour light/dark photoperiods in a growth chamber. For the seedling resprouting / cotyledon latent meristem burying study, ten (out of 240) greenhouse-grown plants were cut at soil level each month for 24 months to determine age of resprouting, and ten (out of 120) greenhouse-grown plants were examined every other month for 24 months to measure height of cotyledon latent meristem above (or below) soil level. All greenhouse plants that were selected each month (or every other month) were measured for height and dry mass (total or above-ground). Herbicide and TNC data were analyzed using a randomized complete block design ANOVA (with study site as blocking factor), with appropriate post hoc tests to

Texas Tech University, Pablo Teveni, August 2017 separate means; Akaike’s Information Criterion (AIC) was used to determine the best-fit model for mortality data. ANOVA for completely randomized design (with appropriate post-hoc tests) was used to analyze germination (daily and cumulative) and monthly change in cotyledon latent meristem height, and probit analysis and linear regression were used to analyze monthly changes in percentage of plants with cotyledon latent meristems at or below soil level and percentage of seedlings resprouting following top removal. Root crown TNC results reveal significant increases (compared to the prior month) during May, August, and December. Increases in TNC from one month to the next resulted from translocation of TNC from aerial portions of the plant into the root crown; in other weedy species foliar application during downward translocation of TNC has increased mortality of treated plants. Herbicide-induced mortality was greatest during the months of October, May, November, and September. The best fit model for mortality was a sixth-order polynomial function of mortality vs. month; when month was removed as a parameter, the best fit model was a quadratic function of mortality vs. soil temperature, combined with a quadratic function of mortality vs. phenology. Although overall germination percentages were the same among temperature treatments, seeds in the 20/10°C temperature treatment germinated more slowly. Seedling resprouting began at 6 months and increased to over 50% at 20 months. Seedling cotyledon latent meristem height had a quadratic relationship with time, with average cotyledon latent meristem height at or below soil level at 22 months of age. Seedling cotyledon latent meristems began to become buried at 7 months, with over 50% becoming buried at 21 months of age. These results indicate two windows in which to chemically treat huisache to achieve high mortality in the coastal plains of South Texas: spring (especially May) and autumn (especially October and November). Besides month (which can have variable environmental conditions), high huisache mortality rates can also be expected when soil temperature at 0.3 m is at or near a peak of 24.5°C, during the full canopy phenological stage. Huisache seeds can be expected to germinate over a wide range of temperatures, with germination rate slowing when high/low temperatures are lowered to 20° and 10° C. Controlling huisache through top-removal methods can result in over 50% of plants

vii

Texas Tech University, Pablo Teveni, August 2017 resprouting when 20 to 22 months of age.

viii

Texas Tech University, Pablo Teveni, August 2017

List of Tables Table Page

1. Single-factor analysis of variance table for dependent variable Huisache Root Crown TNC, across all sites, with fixed factors Phenological Stage and Site (blocking factor)………………………………...... 37

2. Single-factor analysis of variance table for dependent variable Huisache Mortality, across all sites, with fixed factors Herbicide Formulation, Phenological Stage, and Site (blocking factor)...... 39

3. Single-factor analysis of variance table for dependent variable Huisache Root Crown TNC, across all sites, with fixed factors Month*Year and Site (blocking factor).…………………………………………40

4. Single-factor analysis of variance table for dependent variable Huisache Mortality, across all sites, with fixed factors Month* Year and Site (blocking factor).………………………………..………………...40

5. Single-factor analysis of Variance table for dependent variable Huisache Root Crown TNC, across both sites with predominantly sandy loam / sandy clay loam soil (Diebel Ranch and King Ranch), with fixed factors Month*Year and Site (blocking factor)..…………...... 41

6. Single-factor analysis of variance table for dependent variable Huisache Mortality, across both sites with predominantly sandy loam / sandy clay loam soil (Diebel Ranch and King Ranch), with fixed factors Month*Year and Site (blocking factor)……………………...... 41

7. Single factor analysis of variance table for dependent variable Huisache Root Crown TNC, across both sites with predominantly clay soil (Baumgartner Ranch and Welder Refuge), with fixed factors Month*Year and Site (blocking factor)………………………………….42

8. Single-factor analysis of variance table for dependent variable Huisache Mortality, across both sites with predominantly clay soil (Baumgartner Ranch and Welder Refuge), with fixed factors Month*Year and Site (blocking factor).)………………………………….……..42

9. Single-factor analysis of variance table for dependent variable Cumulative Total Germination, with fixed factors Scarification * Temperature ………………………………………………...... 58

Texas Tech University, Pablo Teveni, August 2017

10. Single-factor analysis of variance table for dependent variables Cumulative Germination and Non-cumulative Germination, with fixed factors Temperature * No. of Days (in germination chamber), for Scarified and Non-scarified seeds……….……………………….…………..60

11. Cumulative germination per day*temperature treatment for Scarified huisache seeds. Statistical significance determined using ANOVA with completely randomized design and Duncan’s Multiple Range test…………………………………………………………...... 62

12. Fiducial confidence interval table for dependent variable Seedling Resprouting (%), with covariate plant age (in months), determined using Probit analysis.)…………………………………………....………..……..64

13. Single-factor analysis of variance table for dependent variable Cotyledon Latent Meristem Height (mm), with fixed factor No. of Months (in age)………………………..………………………………………....65

14. Fiducial confidence interval table for dependent variable Cotyledon Latent Meristem Below Soil (%), with covariate plant age (in months), determined using Probit analysis.………………………...…....67

Texas Tech University, Pablo Teveni, August 2017

List of Figures

Figure Page

1. Average total nonstructural carbohydrate (TNC, %) for different phenological stages. Statistical significance determined using ANOVA with randomized complete block design and Duncan’s multiple range test where appropriate. Bars represent ± SE..……………..…….38

2. Average total herbicide-induced huisache mortality (%) for A) Soil Type, and B) Phenological Stage. Statistical significance determined using ANOVA with randomized complete block design and Duncan’s multiple range test where appropriate. Bars represent ± SE...………………………………………………………………….39

3. Average monthly total nonstructural carbohydrate content (TNC, %) and herbicide-induced mortality (%) across all study sites. Statistical significance was determined using ANOVA with randomized complete block design and Fisher’s Protected LSD test for TNC and Duncan’s Multiple Range test for mortality. Bars represent ± SE...………………………………………………………...…..43

4. Average monthly total nonstructural carbohydrate content (TNC, %) and herbicide-induced mortality (%) for predominantly sandy loam / sandy clay loam sites. Statistical significance determined using ANOVA with randomized complete block design and Fisher’s Protected LSD test for TNC and Duncan’s Multiple Range test for mortality. Bars represent ± SE...………………...………………...……44

5. Average monthly total nonstructural carbohydrate content (TNC, %) and herbicide-induced mortality (%) for predominantly clay sites. Statistical significance determined using ANOVA with randomized complete block design and Fisher’s Protected LSD test for TNC and Duncan’s Multiple Range test for mortality. Bars represent ± SE.…………………………………………………...... 45

6. Average daily soil temperature below 0.3 m and daily precipitation for all four study sites, during period of field study (mid-April 2012 to mid-November 2014): A= Baumgartner Ranch, B = Diebel Ranch, C = King Ranch, and D = Welder Wildlife Refuge……………………………………………………………………………46

7. Best-Fit Model (Huisache Mortality vs. Month) for all data, according to AIC (Akaike’s Information Criterion) analysis (AIC = 1169.382, AICc [AIC corrected for small sample sizes] = xi

Texas Tech University, Pablo Teveni, August 2017

1170.015, R2 = 0.4054, Adj. R2 = 0.3854, and n 185)…………………………...49

8. Best fit model for all data (A) Huisache Mortality vs. Soil Temperature at 0.3 m and B) Huisache Mortality vs. Phenology), according to AIC analysis (AIC = 1117.623, AICc = 1117.983, R2 = 0.3123, Adj. R2 = 0.301, and n =173), when the parameter month is removed from analysis…………………………………………...... 50

9. Best fit model for data from predominantly sandy loam / sandy clay loam sites (A) Huisache Mortality vs. Month, B) Huisache Mortality vs. Phenology, and C) Huisache Mortality vs. Delta- TNC [current month TNC – previous month TNC), according to AIC analysis (AIC = 594.377, AICc = 597.028, R2 = 0.537, Adj. R2 = 0.522, and n = 94)……………………….………………..…………….…..52

10. Best fit model for data from predominantly sandy loam / sandy clay loam sites (A) Huisache Mortality vs. Soil Temperature at 0.3 m and B) Huisache Mortality vs. Cumulative 8-week Precipitation), according to AIC analysis (AIC = 490.98, AICc = 491.559, R2 = 0.339, Adj. R2 = 0.31, and n = 74), when the parameter month is removed from analysis………………………...... 53

11. Best fit model for data from predominantly clay sites (A) Huisache Mortality vs. Soil Temperature at 0.3 m and B) Huisache Mortality vs. Delta-TNC), according to AIC analysis (AIC = 511.725, AICc = 512.238, R2 = 0.308, Adj. R2 = 0.282, and n = 83)...…………...... 55

12. Best fit model for full canopy phenological stage data from all sites (Huisache Mortality vs. Soil Temperature), according to AIC analysis (AIC = 749.722, AICc = 749.92, R2 = 0.1023, Adj. R2 = 0.0875, and n = 125).……………………………………………...... 56

13. Best fit model for data from plants in partial defoliation phenological stage from all sites (A) Huisache Mortality vs. Future-Delta-TNC [next month TNC – current TNC] and B) Huisache Mortality vs. 4-week Cumulative Precipitation), according to AIC analysis (AIC = 170.659, AICc = 171.859, R2 = 0.3549, Adj. R2 = 0.2934, and n = 24)…………………………...... 57

14. Cumulative total germination percentages for Scarified and Non- Scarified huisache seeds growing in a germination chamber, under four different temperature regimes (high/low, °C). Statistical significance determined using ANOVA with completely randomized design and Duncan’s Multiple Range test. Bars represent ± SE..……………………………………………………………..……59 xii

Texas Tech University, Pablo Teveni, August 2017

15. Cumulative germination curves for Scarified and Non-Scarified huisache seeds at four different temperature regimes (high/low, °C). Bars represent ± S.E.……………………………………...... 61

16. Huisache seedling cotyledon latent meristem height above soil level, measured every other month for 24 months and curvilinear estimation for cotyledon latent meristem height above soil level vs. month. Statistical significance determined using ANOVA with completely randomized design and Duncan’s Multiple Range test.……………..65

A-1. Volumetric soil water content (vol./vol.) at 0.3 m for all study sites during period of study (mid-April 2012 to mid-November 2014), with field capacity and permanent wilting points for dominant soil types at 0.3 m: A = Baumgartner Ranch, B = Diebel Ranch, C = King Ranch, and D = Welder Wildlife Refuge…………………………………106

A-2. Best fit model for May data from all sites (A) Huisache Mortality vs. Future-Delta-TNC [next month TNC – current TNC] and B) Huisache Mortality vs. 4-week Cumulative Precipitation), according to AIC analysis (AIC = 67.804, AICc = 69.804, R2 = 0.577, Adj. R2 = 0.512, and n = 16)………………………………………….....107

A-3. Best fit model for October data from all sites (Huisache Mortality vs. 4-week Cumulative Precipitation), according to AIC analysis (AIC =83.022, AICc = 83.822, R2 = 0.07, Adj. R2 = 0.012, and n = 18)…………………………………………………………………………...….108

A-4. Best fit model (1 of 2) for November data from all sites (Huisache Mortality vs. Soil AWC (available water content, %)) at 0.3 m, according to AIC analysis (AIC = 80.61, AICc = 81.533, R2 = 0.039, Adj. R2 = -0.029, and n = 16)……………………………………….…..108

A-5. Best fit model (2 of 2) for November data from all sites (Huisache Mortality vs. Soil AWC (available water content, %)) at 0.3 m, according to AIC analysis (AIC = 89.769, AICc = 90.569, R2 = 0.061, Adj. R2 = 0.003, and n = 18)………………………………………..….109

xiii

Texas Tech University, Pablo Teveni, August 2017

Chapter 1

INTRODUCTION

Huisache (Acacia farnesiana [L.] Willd.) is a thorny leguminous shrub or small tree that has aggressively invaded rangelands in southern Texas. This species is also known as Vachellia farnesiana (L.) Wright and Arn., Acacia smallii Isley, A. minuta (M.

E. Jones) R. M. Beauch., Pomponax farnesiana (L.) Raf., and others (Ebinger et al.

2000). Ebinger et al. (2002) divided Acacia farnesiana into three subspecies: A. farnesiana (L.) Willd. subsp. farnesiana, A. f. (L.) Willd. subsp. minuta (M. E. Jones)

Ebinger, Siegler, and Clark, and A. f. (L.) Willd. subsp. pinetorum (F. J. Hermann)

Ebinger, Siegler and Clark, with A. f. subsp. farnesiana comprising the A. smallii phenotype and commonly occurring in southern Texas. Huisache is also known by other common names, including: sweet acacia and aroma (Parrotta 1992).

A. farnesiana is found across the southern United States, from southern Florida to southern California, through Central America and the Caribbean islands to Argentina in

South America (Ebinger et al. 2002). It has naturalized throughout the world in both tropical and subtropical habitats, in locations such as southern Europe (Parrotta 1992;

Tutin 1980), Canary Islands (Kunkel 1976; Naranjo et al. 2009), North Africa (El-Lakany

1987), Nigeria (Gill et al. 1986), India (Joshi 1983), China (Xiangyun 2002), Australia

(Batianoff and Butler 2002), Hawaii (Stone et al. 1992), and other islands in the Pacific

Ocean (Meyer 2000). Huisache can grow in soils of various textures (Lonard et al. 1991;

Pennington and Sarukhan 1968), of various pH levels (optimum range between pH 6 to 8;

Scifres 1974), and under annual precipitation levels ranging from 40 cm (Badi 1967) to over 100 cm (personal observation).

Texas Tech University, Pablo Teveni, August 2017

This species is considered to be a rangeland pest in coastal South Texas because heavy infestations reduce grass production (Scifres et al. 1982). It also resprouts quickly following disturbance (Dacy and Fulbright 2009); produces seeds abundantly (Scifres

1974); and is difficult to control by mechanical, chemical, or pyric means (Box et al.

1967; Dacy and Fulbright 2009; Rasmussen et al. 1983, Powell et al. 1972; Meyer et al.

1976). Huisache has the ability to grow in non-grazed grasslands and prairies (Meyer and Bovey 1982) and grows more quickly than most other woody species in South Texas

(personal observation).

Despite its problematic status as an invasive shrubby species, it is also worth noting that huisache has both agricultural and industrial economic value in other parts of the world. For example in France and other parts of southern Europe and the

Mediterranean Basin, huisache is cultivated, and the flowers are harvested to produce essential oils, to make perfumes and other products (El-Gamassy and Rofaeel 1975;

Lawrence 1984; Parrotta 1992). However, the use of huisache for agricultural and industrial purposes is not known to have ever occurred in Texas, or anywhere else in the

United States of America.

Huisache History in South Texas

In South Texas, huisache thrives in locations with low soil fertility and high light intensity (Fulbright et al. 1997; Van Auken and Bush 1985), most dominantly in poorly drained shrublands and riparian woodlands (McLendon 1991). Although drought tolerant

(Scifres 1974), huisache growth and proliferation appears to be increased on wet sites and during wet periods (Mutz et al. 1978).

Texas Tech University, Pablo Teveni, August 2017

At the turn of the twentieth century, huisache was only found within about 100 miles inland from the Gulf coast, normally growing in clay or “heavy and firm” sandy soils, especially along streams and around ponds, while lighter sandy soils were mostly open prairie (Bailey 1905; Schmidly 2003). Previously, from the 1500’s to the 1800’s,

South Texas had been similarly described as mostly grassy plains and prairies, with scattered shrubs, trees, shrublands, thickets, and woodlands, especially along streams and rivers (Bartlett 1854; Berlandier 1980; Duaine 1971; Forrestal 1934; Fulbright 1991;

Céliz 1935). These observations were supported by Carbon-13 isotope analysis of soils beneath herbaceous and woody plant cluster components in South Texas savannas

(Archer 1990; Tieszen and Archer 1990), by models based on growth of individual woody plants and clusters (Archer 1989), by models based on landscape transition

(Scanlan and Archer 1991), and by tree-ring analysis (Boutton et al 1998).

The relative balance in abundance and distribution between herbaceous and woody vegetation types in savannas is considered to be a result of complex relationships among soils, climate, geomorphology, topography, fire, and herbivory (Backéus 1992;

Scholes and Archer 1997; Walker 1987). Human activities have also been important in the history of many savannas (Scholes and Archer 1997). Before European settlement in the Americas, fires probably burned South Texas rangelands every five to thirty years

(Wright and Bailey 1982), often caused by Native American tribes (Berlandier 1834;

Newcomb 1961; Nuñez Cabeza de Vaca 1871). The suppression of this wildfire regime, along with historic overgrazing of herbaceous cover by livestock, is believed to have led to an increase in the spread of huisache and other brush species, at the expense of herbaceous grassland species (Archer et al. 1988; Archer 1994; Fulbright 1991; Fulbright

Texas Tech University, Pablo Teveni, August 2017

2001; Johnston 1963). This is supported by recommendations for prescribed burns every two to five years to decrease woody vegetation canopy cover, density and frequency in

South Texas rangelands (Drawe 1980; Rasmussen et al. 1983; Ruthven et al. 2003).

Studies that show huisache growth suppression by both C3 and C4 grasses growing in proximity, with little reduction in growth of the grasses also support this (Cohn et al.

1989; Van Auken and Bush 1990; Van Auken 1994).

Overgrazing can increase woody plant densities in several ways: 1) preferential removal of grass biomass (release of shrub seedlings from competition), 2) increased stress (reducing competitive ability) of grazed plants, 3) shrub seed dispersal, 4) reduction of fine fuels for fire, 5) shifting the herbaceous plant composition into one that is less competitive against woody species, 6) soil compaction (reduced infiltration), and

7) soil erosion, thus changing soil properties to favor woody species (Archer 1994;

Archer 1995a; Archer 1996). However, huisache has been shown to have the ability to become established even in non-grazed rangelands (Meyer and Bovey 1982), and brush cover can increase rapidly in savannas free from fire, grazing, and browsing (Brown and

Archer 1989; San José and Fariñas 1991). Huisache also grows more quickly than most co-occurring brush species, including mesquite (personal observation). Therefore, promotion of a more natural wildfire regime and prevention of overgrazing are probably both required to keep huisache from further invading grasslands. The advent of large- scale fencing during the late 1800’s contributed to an increase in overgrazing by limiting the movements of livestock (Archer 1994; Bogusch 1952), and the historic extirpation, decimation, and other alterations of native herbivore populations during recent centuries might also have helped increase the spread of woody species (Archer 1994; Archer

Texas Tech University, Pablo Teveni, August 2017

1996). The consumption of huisache pods by livestock and subsequent dispersal of seeds to new locations also helped increase the range and density of huisache (Parrotta 1992).

Climate change is another factor believed to have contributed to the increase in brush range and density in South Texas (Fulbright 1991). Increased atmospheric CO2, a global concern in recent decades, has been shown to augment nitrogen fixation and biomass accumulation of huisache, which increases its competitiveness over non- nitrogen-fixing plants growing in low-nitrogen soils (Muttiah and Johnson 1999; Polley et al. 1997, Tischler et al. 2004). In particular, increased atmospheric CO2 since the start of the Industrial Revolution is believed to have increased the competitive advantage of C3 shrubs (such as huisache) over C4 grasses, which dominate Texas prairies (McCarron and

Knapp 2001) and grasslands and savannas in the southwestern United States (Idso 1992;

Johnson et al. 1993; Mayeux et al. 1991). Increased atmospheric CO2 may also increase the spread of trees in savannas by increasing the rate of resprouting and the rate of stem elongation, thus decreasing tree mortality from fire (Bond and Midgley 2000). In addition, severe and frequent droughts (in combination with ensuing pluvial periods) are believed to have influenced or mediated the spread of shrublands into grasslands (Archer

1990; Fulbright and Bryant 2003; Harrington 1991; Herbel et al. 1972), and in fact, some grassland studies have shown drought to have a greater negative effect upon grass plant basal area than level of grazing (Branson 1985; Clarkson and Lee 1988).

Dynamic changes of climate during the past several millennia are evidenced by pollen analyses in southwest Texas (near Del Rio) that indicate a wetter climate existing

10000 years ago, with a dryer climate following 4000 to 1000 years ago, increasing woody vegetation adapted to arid and semiarid environments (Johnson 1963; Hester

Texas Tech University, Pablo Teveni, August 2017

1980). Glacial ice cores and the fossil record indicate significant instability in climate and vegetation in Southwestern North America over time scales stretching from decades to millennia (Archer 1994). Studies have shown that perennial plants, especially deeply- rooted and long-lived woody plants, can persist for decades or centuries through climatic changes resulting in conditions (e.g. temperature and soil moisture) that do not closely resemble those that were present during the relatively narrow window of seed germination and establishment (Cole 1985; Lewin 1985). Therefore, it is possible that grasslands and savannas that became established under the influence of previous climates are only somewhat suited for current or recent climates (Archer 1994; Neilson 1986).

Huisache, however, is well adapted to a wide variety of different climates, as evidenced by its world-wide distribution and phenological plasticity (Parrotta 1992; Naranjo et al.

2009).

By the early 1960’s huisache had infested 1.1 million ha of rangeland in Texas, primarily in the Coastal Prairies and South Texas Plains (Smith and Rechenthin 1964).

According to vegetation surveys, huisache canopy cover increased by 48% in Texas from

1964 to 1982 (USDA 1985). Though as yet not documented in the literature, there is widespread concern that huisache has continued to expand in range and density during the last 35 years.

Effect of Huisache on Rangelands

Several attributes of huisache contribute to its status as a pest plant species. On the Coastal Prairies of South Texas, annual production of grasses decreased with an increase in huisache canopy cover beyond 32% (Scifres et al. 1982). This is because

Texas Tech University, Pablo Teveni, August 2017 brush species compete with grasses and other herbaceous plants for sunlight, water, space, and soil nutrients (Bade 1991, Scholes and Archer 1997), and because woody canopy interception and accumulated litter below the canopies can significantly reduce soil moisture (Návar and Bryan 1990; Scholes and Archer 1997). Huisache and other woody plants with deep taproots can also extend their roots deeper than the root zone of grasses, enabling them to access subsurface moisture unavailable to grasses, especially in coarse-textured soil (Brown and Archer 1990; Knoop and Walker 1985; Walker et al.

1981; Walker and Noy-Meir 1982; Walter 1972). Woody plants can also out-compete grasses under conditions of soil infertility, shallow soil depth, or soil erosion (Archer

1994), and woody plants can change soil nutrient distribution to favor the proliferation of woody species over herbaceous species (Archer 1996; Schlesinger et al. 1990). C4 grasses, which are dominant over C3 grasses in south Texas rangelands (as well as in many other subtropical and temperate grasslands, prairies, and savannas; Risser 1981), do not grow as well as C3 grasses beneath shrub canopies because C4 grasses require higher temperatures, grow better at higher levels of irradiance (Kanai and Black 1972), and grow less during the autumn, winter, and spring (when many shrub canopies are senescent, dormant, or in initiating growth, allowing greater light infiltration) than C3 grasses (Parton and Risser 1976; Teeri and Stowe 1976). Therefore, shrub encroachment has a more negative effect upon C4 grasses than upon C3 grasses.

Cattle on the Coastal Prairies of South Texas have little preference for browsing shrubby species (Drawe and Box 1968). Heavy brush densities add difficulty to working and gathering livestock and can decrease forage and habitat quality for wildlife species

(Welch 1991; Welch undated). Huisache is also poor browse for goat production (Welch

Texas Tech University, Pablo Teveni, August 2017

1991). Through increased canopy interception and reduced soil infiltration, heavy brush infestations might also reduce the amount of groundwater and surface water in rangeland watersheds, and the resulting reduction in grass cover can cause increased soil erosion, lowering the quality of surface water runoff (Sosebee 1987; Hibbert 1983; Jensen 1988;

Kelton 1975; Scholes and Archer 1997; Welch 1991). Savanna trees are most likely to reduce herbaceous production in rangelands with high precipitation and high tree density

(Beale 1973; Burrows et al. 1988; Pieper 1990; Sandford et al. 1982). High levels of livestock grazing might also reduce herbaceous production beneath savanna trees through increased livestock use of shaded locations (Belsky et al. 1993b). Also, studies with other nitrogen-fixing species has shown that changing the nitrogen balance of rangeland soils can cause increases in other weedy and invasive species (Maron and Connors 1996).

Because huisache generally grows faster than most other woody rangeland plant species in South Texas (personal observation) and because huisache grows well in very heavy soils (personal observation) while most other brush species grow best in soils without argillic horizons (Archer 1995b; Barnes and Archer 1999), it appears likely that huisache cannot only outcompete other brush species when growing in heavier soils but can also drastically increase density on rangelands with heavy soils, locations which otherwise

(without the presence of huisache) might not be particularly prone to large-scale brush infestations. In addition, large, thick stands of trees or brush (i.e. more than 371 huisache trees per hectare (150 trees per acre) in the Gulf Coastal Prairies) can prevent prescribed burns from achieving complete coverage of certain areas, due to the discontinuity of fine fuels below shrub and tree canopies (McLendon 1991; Hamilton undated; Drawe 1980;

Box et al. 1967). Finally, huisache is a preferred nesting and roosting species for the

Texas Tech University, Pablo Teveni, August 2017 great-tailed grackle (Quiscalus mexicanus), a nuisance avian species in South Texas, particularly during the breeding season (Hobbs and Leon 1987).

Conversely, huisache can confer some benefits to rangelands. Non-treated areas of mixed brush species, including huisache, in South Texas provided better white-tailed deer (Odocoileus virginianus) habitat and forage, compared with nearby root plowed and re-seeded grassy areas, especially during drought, frost, and overstocking (Davis and

Winkler 1968; Fulbright and Taylor 2001). Huisache in the Coastal Prairies of South

Texas can have a moderate to high forage quality rating for white-tailed deer, particularly on clay soils during summer, autumn, and winter (Drawe and Box 1968), and huisache can have relatively high levels of crude protein during the spring and summer months, compared with other browse species (Wright et al. 2002). Although it is of low browse preference, huisache can provide a large percentage of mast for white-tailed deer in the summer (approximately 20%) and autumn (approximately 50%), without negatively affecting population and nutritional status, especially in locations where other browse species have limited occurrence (Ruthven et al. 1994). Maintenance of 40 to 60 % brush cover has been shown to provide the most successful conditions for white-tail deer management (Richardson 1989; Steuter and Wright 1980). Likewise, despite its low preference, cattle can browse brush, including huisache, for 20% of their diet in the winter and 10% of their diet in the summer (Everitt et al. 1981), and cattle usually overwinter in better condition on brushy pastures than on grassy pastures, possibly due to more protein in browse and forage in the former than in the latter (Hanselka undated).

Increasing huisache canopy cover up to 25% can increase Texas wintergrass (Stipa leucotricha), an important cool-season forage for cattle, and increasing huisache canopy

Texas Tech University, Pablo Teveni, August 2017 cover can also improve winter forage by reducing winter weathering of the standing crop of warm-season grasses (Scifres et al. 1982). Similarly, a mixed-brush/bristlegrass community (including huisache) on the Gulf Coastal Prairie was found to produce more grass between the months of November and January, compared with prickly pear/shortgrass, bunchgrass/annual forb, or mesquite/buffalograss communities, yet all communities (except prickly pear/shortgrass) yielded similar amounts of herbage production over the annual period (Box 1960).

Huisache can also add organic matter and plant-available nitrogen to rangeland soils through nitrogen-fixation by Rhizobium bacteria in root nodules, especially in locations with relatively low soil nitrogen (Herrera-Arreola et al. 2007; Zitzer et al.

1996). In arid and semi-arid areas, huisache and other leguminous woody species can actually create islands of soil fertility beneath their canopies, compared with soil in interspaces (Geesing et al. 2000; Reyes-Reyes et al. 2003; Reyes-Reyes et al. 2002;

Virginia and Jarrell 1983) through reduced solar irradiance (resulting in lower soil temperatures, reduced plant evaporative demand, increased soil moisture, and increased plant water-use efficiency), canopy trapping of nutrient-rich dust, hydraulic lift, increased animal use (increased defecation and burrowing), and enhanced microbial biomass

(Amundson et al. 1995; Belsky et al. 1989; Caldwell and Richards 1989; Dawson 1993;

Mordelet and Menaut 1995; Virginia 1986; Weltzin and Coughenour 1990). This beneficial effect is especially evident in situations of high abiotic stress (Bertness and

Callaway 1994), such as drought (Barnes and Archer 1999; Bernhard-Reversat 1982).

The removal of mesquite can cause decreases of available soil nitrogen, phosphorous, potassium, sulfur, and iron (Frías-Hernández et al. 1999; Tiedemann and Klemmedson

Texas Tech University, Pablo Teveni, August 2017

1986); removal of huisache might cause similar declines in soil fertility. Deep roots (> 1 m) might also allow woody savanna species to capture nutrients below the herbaceous root zone, reducing nutrient leaching, recycling those nutrients, and keeping them in the ecosystem (Belsky et al. 1989). The beneficial effects of savanna trees upon forage productivity are most likely to be realized in tropical and subtropical rangelands where soil fertility is moderate, annual precipitation is low, tree density is low, and grazing intensity is low or moderate (Kennard and Walker 1973; Knoop and Walker 1985;

Burrows et al. 1990; Belsky et al. 1993a; Belsky et al. 1993b; Belsky 1994).

Although “thicketization” (i.e. transformation of grasslands and prairies into shrublands or woodlands) of mesic grasslands and savannas can result in socioeconomic degradation due to decreased livestock carrying capacity of rangelands, it does not necessarily cause ecological degradation if levels of productivity, biodiversity, and other ecological factors are not diminished; therefore land management could potentially re- focus efforts on other rangeland products or services that hold socioeconomic value, such as lease hunting or ecotourism (Archer et al. 2000). Long-term increases in carbon levels in vegetation and associated soils as grasslands transform into woodlands indicate that they serve as carbon sinks (Archer et al 2000; Ciais et al. 1995), potentially a socially valuable carbon sequestration service (Glenn et al. 1992). Huisache can also provide nesting sites for some raptor species (Actkinson et al. 2007) and for the plains chachalaca

(Ortalis vetula mccalli) of the Lower Rio Grande Valley (Marion and Fleetwood 1978), and can also provide cover for Texas bobwhite quail (Colinus virginianus texanus), especially during summer, autumn, and winter (Lehmann undated). Huisache also provides forage for small mammals, mast for small mammals, songbirds, and game birds,

Texas Tech University, Pablo Teveni, August 2017 cover for small mammals and songbirds, and nectar and pollen for butterflies and other pollinators (Lyons et al. 1999; Davis 1934).

Huisache Control Methods

Brush treatment of huisache can be done either for reclamation or for maintenance. Reclamation is usually done when brush densities are relatively high and is intended to significantly reduce brush densities, while maintenance is usually done when brush densities are relatively low and is intended to maintain brush densities at or near their current level (Welch undated). Because brush control can be a costly investment for land managers, it is recommended that reclamation efforts be followed up by periodic, long-term maintenance efforts (Fulbright 1991; Welch undated, Hamilton undated,

Whitson and Scifres 1980). Whereas brush eradication was widely the goal for land managers prior to the mid-1970’s, brush management, in accordance with land use, has generally become the goal since then, as brush eradication was revealed to be unfeasible or impossible, and as the uses for brush (especially as wildlife forage and habitat) became more understood and accepted (Welch 1991; Fulbright 1991).

Removal of huisache for increased forage production can be problematic. It is difficult to control by burning (Box et al. 1967; Dacy and Fulbright 2009; Rasmussen et al. 1983) or by mechanical treatment (Powell et al. 1972) because it re-sprouts quickly

(Dacy and Fulbright 2009) from deep stem tissues (Bontrager et al. 1979), produces seeds prolifically (Scifres 1974), and the seeds readily germinate in disturbed soils (Mutz et al.

1978) and are easily dispersed by cattle herbivory (Parrotta 1992). Soil disturbance caused by the removal of underground portions of huisache can create conditions

Texas Tech University, Pablo Teveni, August 2017 favorable for the emergence of huisache seedlings (Mutz et al. 1978).

Although controlled burning can decrease huisache canopy cover by 90% or more over the short term, all or nearly all plants over 1 m tall survive, and plants burned at ~ 2 m tall mostly return to their original height after 2 years, suggesting that although fire might help limit the spread of huisache, fire alone probably cannot reduce huisache infestations (Rasmussen et al. 1983). Burning at different times of the year and burning over different soil types can have differing effects on the grass and forb community. For example, in the Coastal Prairies of South Texas, burning in December increased forb production on sandy but not on clay soils; burning in January increased grass production on clay soils but not on sand; and burning in February increased grass production on soils of both sand and clay (Hansmire et al. 1988). Burning huisache can also temporarily improve management for white-tailed deer by increasing the sprouting of succulent new growth, which is higher in crude protein and phosphorous than in non-treated plants

(Rasmussen et al. 1983; Hanselka et al. 2004). Low-severity burning of woodland savannas can temporarily increase soil microbial carbon, soil nitrate content, and potential denitrification activity; however, these effects may only last for between three and seven months (Andersson et al. 2004).

Mechanical treatments such as shredding, chaining, cabling, roller chopping, scalping, and root plowing can actually increase huisache densities and make the land less suitable for wildlife and livestock (Box and Powell 1965; Fulbright 1991; Fulbright and Taylor 2001; Mutz et al. 1978). For example, compared with non-treated sites nearby, root plowing in the eastern South Texas Plains caused huisache density to increase seven-fold after 17 years, while brush species diversity declined and some brush

Texas Tech University, Pablo Teveni, August 2017 species preferred by white-tailed deer disappeared entirely (Ruthven et al. 1993). This might have been due to an abundance of shrub seedlings following root plowing (Allison and Rechenthin 1956). Some mechanical treatments can also spread and increase problem species, such as prickly pear (Opuntia spp.) (Allison and Rechenthin 1956;

Fulbright 1991). In addition, some mechanical treatments increase brush density by changing the growth habit of brush species from single or few stems to multiple stems

(Fulbright 1991). However, some mechanical treatments can temporarily improve forage management by causing shrubs and trees to sprout abundant new growth, increasing preference values and forage ratings of brush species for deer and cattle, including increasing preference values and crude protein percentages of huisache browse (Powell and Box 1966). Clearing of woody plants can also benefit herbaceous plant growth through the enrichment of soil by residue decomposition, which can increase herbaceous production for over 13 years following woody plant removal (Tiedemann and

Klemmedson 1986; Klemmedson and Tiedemann 1986).

Mechanical treatment of brush can be complicated by wet soils, clay soils

(especially dry clay soils), and shallow, rocky soils, and treatments are most effective when soil is moist but not wet (Welch undated; Hamilton undated; Hoffman 1981; Drawe

1980; Fisher et al. 1973). Mechanically treated areas with poor range condition (10 to

15% of desirable forage potential) also need to be reseeded with desirable forage species soon afterward, in order to achieve optimum forage production, and forage production will be also largely dependent upon sufficient soil moisture, soil fertility, and soil depth

(Welch undated). Of the woody plants in South Texas that re-sprout following top removal, huisache sprouts appear to elongate most quickly, growing up to 1.19 m after

Texas Tech University, Pablo Teveni, August 2017 only five months following shredding (Powell et al. 1972). This allows it to quickly shade out other range plants. Power grubbing huisache at depths of 20 cm on sandy loam soils on the Coastal Prairie killed only 79% of plants, because of the presence of stem tissues deep below the ground (> 20 cm; Bontrager et al. 1979). Mechanical control methods and fire can be somewhat more effective when used together than when either is used alone (Box et al. 1967; Drawe 1980). However, the inadequacy of treatment techniques requires continuous brush management of huisache and other shrubby species for years or decades, as trees and brush may re-colonize within 3 to 10 years and eventually return the site to its pre-treatment condition (Heitschmidt et al. 1986).

Biological control of huisache is minimal, if existent. Huisache twig girdler

(Oncideres spp.) is an insect that often removes huisache branches; however, this species usually does not girdle the plant close enough to the root crown to prevent resprouting

(personal observation). Bruchid beetles (Mimosestes spp.) often consume huisache seeds

(Naranjo et al. 2009). In other savanna and grassland ecosystems, browsers, herbivorous rodents, and insects help to reduce densities or control the spread of woody species (Crisp and Lange 1976; Pellew 1983; Berdowski 1987; Yeaton 1988; Cantor and Whitham

1989; McPherson 1993; Weltzin et al. 1997).

Several different herbicides can control huisache when applied to individual plants (Bovey and Meyer 1989, Hanselka et al. 2010), but at large scales, such treatments become cost prohibitive in time and resources. When aerially applied, many herbicides can achieve high rates of canopy reduction, but few cause predictable and complete mortality, although idiosyncratic examples of mortality can be found in the literature.

For example, picloram (4-amino-3,5,6-trichloropicolinic acid); 2,4,5-T [(2,4,5-

Texas Tech University, Pablo Teveni, August 2017 trichlorophenoxy)acetic acid]; bromacil (5-bromo-3-sec-butyl-6-methyluracil); isocil (5- bromo-3-isopropyl-G-methyluracil); and clopyralid (3,6-dichloro-2-pyridinecarboxylic acid) have all been found to kill huisache under some conditions, though success remains unpredictable (Bovey et al. 1967, Bovey et al. 1970, Bovey et al. 1972, Bovey and Meyer

1989, Bovey et al. 1990). Application of herbicides has also raised concern about negative impacts upon groundwater resources and endangered species (McGinty 1991), and non-susceptible understory weedy species have been known to increase following chemical control of other woody species, such as oak (Quercus spp.) and mesquite

(Prosopis spp.) (Hamilton 1991). Also, different herbicides have different control rates against different species, complicating management decisions when multiple problem species are present, as is the case across many South Texas rangelands (Allison and

Rechenthin 1956). The efficacy of herbicide treatment can vary according to soil water status, soil type, and weather variables (Fisher et al. 1956; Dahl et al. 1971; Scifres 1980).

In mesquite, herbicide-induced mortality was found to vary according to age of treated resprouts, though differences were greater among different herbicide formulations and among different months of treatment (Beck et al. 1975).

In order to achieve optimum increases in forage production, range condition prior to broadcast herbicide treatment should be at least fair (with a similarity index of 25 to

50% of desirable forage potential), because treated areas with low poor range condition

(10 to 15% of desirable forage potential) will naturally revegetate very slowly (Welch undated). Brush to be sprayed should have full canopies of mature leaves with high subsoil moisture and temperatures between 21.1 and 23.9 degrees Celsius at 30 to 46 cm below ground, and non-treated smaller huisache plants (under 1.5 m) are more easily

Texas Tech University, Pablo Teveni, August 2017 controlled with herbicide than more mature (above 1.5 m) or previously mechanically treated shrubs (Welch undated; Scifres 1980; Hanselka 1991). Plant physiology and rate of herbicide application will also affect the effectiveness of herbicide applications

(Hamilton undated).

When huisache was treated with picloram, defoliated plants were much less affected than non-defoliated plants, and sprayed plants needed to retain their leaves for at least 24 hours to achieve the maximum kill rate in the greenhouse (Bovey et al. 1967), yet

3 days was required for plants in the field to absorb enough herbicide to achieve the maximum kill rate (Meyer et al. 1972). However, compared with “no rain” control treatments, field-grown huisache treated with picloram at 1.1 kg/ha had decreased mortality following simulated rainfall four hours after treatment in late October but not in early July (Bovey et al. 1990), suggesting that foliar-applied picloram may be much more readily absorbed in warmer weather. The above studies indicate that to achieve optimal control rates, huisache should be ideally be sprayed with picloram when a sufficient amount of foliage is present and at least three days before predicted rainfall, unless warm weather allows for faster absorption.

Huisache treated with picloram at 0.367 kg/ha (2 lbs./acre)had high rates (90% or greater) of mortality when sprayed in May or September (Meyer et al. 1976) and had high rates (84% or greater) of canopy reduction 2 years following treatment when sprayed in

May, June, July, or October (Bovey et al. 1970). However, plants treated in April had no canopy reduction 2 years following treatment, and picloram used at 0.184 kg/ha (1 lb./acre) was as effective or nearly as effective in controlling huisache as using 0.367 or

0.551 kg/ha (2 or 3 lbs./acre), when sprayed in October (Bovey et al. 1970). Out of seven

Texas Tech University, Pablo Teveni, August 2017 treatments (not including control), close mowing (3 to 5 cm), 2,4-D (1.1 kg/ha), and picloram (1.1 kg/ha) reduced the number of huisache seedlings from seeds sown three years prior, compared with the untreated control, but only picloram reduced the number of huisache seedlings after 5 years (Meyer and Bovey 1982). More recently, Hanselka and Lyons (2010) have recommended treated huisache with foliar spray of a mixture of picloram and 2,4-D from between cessation of late summer growth to the decrease of soil temperature below 23.9 °C (usually from September through November). No herbicide studies on huisache are known to have used aminocyclopyrachlor or a mixture of 2,4-D, picloram, and aminopyralid, and the comparative effect of foliar-applied herbicide for every month of the year has not been fully investigated. The effect of phenological stage upon huisache herbicide susceptibility has also not been adequately studied.

Revegetation measures may be necessary following successful control of woody species, especially in areas where brush density is high enough to have depleted or significantly reduced the quantity of propagules of desirable herbaceous forage species from the seedbed (Hanselka et al. 1996; Bedunah and Sosebee 1984; Allred 1949).

Revegetation can also help the herbaceous plant species component compete with resprouting or newly-sprouting huisache and other woody brush species (Bush and Van

Auken 1990; Cohn et al. 1989; Van Auken and Bush 1990; Van Auken and Bush 1991;

Van Auken and Bush 1988; Allred 1949).

Total Nonstructural Carbohydrates and Herbicide Susceptibility

In South Texas, huisache flowers from late January or February to March or

April; new leaves begin to grow in February; fruits develop during late March, April and,

Texas Tech University, Pablo Teveni, August 2017

May; and fruits ripen in late May or June (Eddy and Judd 2003; Everitt and Drawe 1993;

Jones 1975; Taylor et al. 1997; Vora 1990). Huisache flowering has been found to be significantly positively correlated with precipitation, with more flowering, and possibly prolonging of flowering season, with increasing precipitation; also, percent fruiting in huisache has been found to be significantly positively correlated with air temperature

(Eddy and Judd 2003).

Studies with other invasive plants, such as mesquite (Prosopis glandulosa Torr.) and broom snakeweed (Gutierrezia sarothrae (Pursh) Britton and Rusby), have shown that measurement of total non-structural carbohydrates (TNC) present in the stem base or root crown is a good predictor of plant susceptibility to herbicide control (Wan and

Sosebee 1990; Sosebee and Dahl 1991; Sosebee 2004). Also, mesquite treated with foliar-applied herbicides were only root-killed when sprayed while soil temperatures were at least 21 to 24 degrees Celsius at 30 to 46 cm below the soil surface, with increasing mortality at warmer temperatures (Sosebee 2004; Dahl et al. 1971). Although no such studies have been performed for huisache, it might also be a species in which the timing for the best control treatments coincide with TNC movement and soil temperature.

Thus, an improved understanding of the environmental physiology of huisache during its different phenological stages may improve the efficacy of control efforts.

In the Lower Rio Grande Valley of Texas, huisache flowers in February, with immature fruit appearing in late March and mature fruit appearing in late May (Eddy and

Judd 2003). In mesquite, which is a co-occurring member of the same family

(Leguminoseae), root TNC is highest (25%) at bud break and decreases with the production of leaves and flowers to a low point around 12%; mesquite root TNC then

Texas Tech University, Pablo Teveni, August 2017 increases with the formation of pods to 20%, lowers to 15% with pod elongation, and rises to 22% with pod maturation (Dahl and Sosebee 1984). Because mesquite and other species are most effectively controlled when carbohydrate translocation into the root crown is increasing, huisache might also respond best to treatment when root TNC is increasing.

Huisache Seed Germination

Although Vora (1989) found that highest germination percentages of huisache seed (97%) occurred following 120 minutes of sulfuric acid (18.4 M) scarification, other researchers have found ideal levels of scarification to be achieved after immersion in concentrated sulfuric acid for 45 to 60 minutes (Scifres 1974), or 20 minutes (Bush

2008). This suggests that large variability exists between seeds of different provenances and also perhaps between different years. Scifres (1974) found the highest germination rates of huisache seed to occur at 30°C. However, germination rates of huisache seeds under diurnally-fluctuating temperatures and under an alternating light/dark cycle (which more close natural conditions) have not been well studied.

Ninety-seven percent of huisache seeds with teguments scarified by sandpaper germinated (Jurado et al. 2000). Scarification by sandpaper can yield germination results as high as 98% (Gill et al. 1986). These studies suggests that scarification with sandpaper might be equivalent (or nearly so) to scarification with concentrated sulfuric acid.

Ideal sowing depths for huisache seed are considered to be between 2 and 4 cm

(Gill et al. 1986). Zisner (1999) found that huisache seeds scarified with a file and grown

Texas Tech University, Pablo Teveni, August 2017 in a greenhouse with bottom heat (22 to 24 degrees C) germinated at 5 days, on average, while Jurado et al. (2000) found that 50% of huisache seeds scarified with sandpaper and grown in a germination chamber at 28 degrees C germinated after 3 days.

Huisache Seedling Resprouting and Cotyledon Latent Meristem Movement

Huisache is well known for its ability to regenerate from growing points below the ground level (Bontrager et al. 1979; Dacy and Fulbright 2009; Rasmussen et al.

1983). Research has suggested that huisache seedlings reemerge more quickly following top removal when soil moisture is not limiting either prior to or following top removal by cutting (Powell et al. 1972) or burning (Rasmussen et al. 1983). Also, resprouting rates were fastest following burning in April and February, but slowest following burning in

June and August (Rasmussen et al. 1983). However, the age and height at which huisache seedlings will regenerate when cut at ground level is not well characterized, even though such knowledge would allow land managers better ability to control huisache when the aboveground portions of the plant can be physically removed to achieve complete mortality. Bovey and Meyer (1974) found that 76% of huisache seedlings survived top removal at eight weeks old, compared with 22% survival at four weeks old; however, in that study, huisache tops were removed within 5 mm of soil level

(rather than at soil level) and there is no description of seed planting depth (which may have affected cotyledon latent meristem positioning following emergence). The age at which seedling huisache cotyledon latent meristems become buried below the soil surface is not well understood. Understanding the depth of the growing points, responsible for resprouting, for specific size and age classes of huisache will allow better knowledge of

Texas Tech University, Pablo Teveni, August 2017 when to treat emerging seedlings, as well as how deep below the soil surface older seedlings should be treated by mechanical means.

Objectives

The objectives of the present research were to understand how root crown TNC, soil texture and temperature, soil water content, precipitation, time of year (i.e. month), and plant phenology influence chemical control of huisache, and to better understand huisache seed germination requirements and seedling response to top-removal. These objectives are met by three separate studies, detailed in the Material and Methods section below. The results of these studies will provide information about how control of huisache can be improved to allow enhancement of herbaceous production, improvement of habitat for livestock and wildlife, and decreased use of herbicides to achieve desired effects.

Texas Tech University, Pablo Teveni, August 2017

Chapter 2

MATERIALS AND METHODS

Study 1: Effective Herbicide Control of Huisache

Description of Study Sites

This study was conducted at four different sites in south Texas, to encompass differences found within the range of huisache, including differences in soil type and precipitation. Sites were selected in: Victoria County (N28o 53’ 03.7”, W96o 57’ 07.1”;

Diebel Cattle Co.), Refugio County (N28o 29’ 23.2”, W97o 08’ 12.2”; Baumgartner

Ranch of Martin O’Connor Cattle Co.), San Patricio County (N28o 06’ 51.7”, W97o 22’

33.8”; Welder Wildlife Refuge), and Kleberg County (N27o 35’ 57.1”, W98o 02’ 01.6”;

King Ranch). All sites were located on the Western Gulf Coastal Plain ecoregion

(Griffith et al. 2004), in which the climax vegetation community is described as a savanna, thorn woodland, grassland, or prairie (Crosswhite 1980; Archer 1990). The ecoregion is also described as the Mesquite-Acacia section of the Prairie Brushland province (Bailey 1980).

The site in Victoria County was characterized by soils that are sandy loam and sandy clay loam at the surface, with sandy clay loam, clay loam, and sandy clay in the subsurface (USDA 2017). The Nada-Cieno complex covered 54% of the area (21.5 ha)

(Nada = sandy loam above 20 cm depth and sandy clay loam between 20 and 203 cm;

Cieno = sandy clay loam above 15 cm, clay loam between 15 and 130 cm, and sandy clay loam between 130 and 203 cm; USDA 2017). Nada soils make up 70% of the Nada-

Cieno complex, while Cieno soils comprise 20% of the soil complex, and the remaining

10% consists of other soil types (U.C. Davis 2017). Telferner fine sandy loam covered

Texas Tech University, Pablo Teveni, August 2017

46% of the area (18.6 ha; fine sandy loam above 41 cm, sandy clay between 41 and 61 cm, and sandy clay loam between 61 and 203 cm; USDA 2017). Nada, Cieno, and

Telferner soils are all poorly-drained soils with very slow permeability and medium to high soil available water capacity, and all three soils experience seasonal wet/drought cycles (Miller 1982). Ecological sites include Claypan Prairie 28-44” PZ, Lowland 35-

56”, and Loamy Prairie 28-40”, and the major land resource area is Gulf Coast Prairies

(USDA 2017). The average annual precipitation at Victoria (10 km SSW of study site) is

104.70 cm (U.S. Climate Data 2017). This site was located in the Northern Humid Gulf

Coastal Prairies sub-ecoregion of the Western Gulf Coastal Plain ecoregion (Griffith et al. 2004). Although the potential climax vegetation is true prairie or wet prairie, the presence of invasive woody species such as huisache and macartney rose indicates continuous heavy grazing in the past (Miller 1982). This site has been used for growing rice in the past (Pers. comm. S. Deiss 2014). The current vegetation on parts of the site can be considered a hackberry-huisache woodland association (McLendon 1991), but much of the site is shortgrass/huisache prairie (Hanselka 2009), wooded prairie community (Carter et al. 2009) or huisache / Chinese tallow / sesbania / sedges / shortgrass prairie community (Hanselka and Reinke 2009). Dominant species included: common broomweed (Amphiachyris dracunculoides), woolly croton (Croton capitatus var. lindheimeri), bitter sneezeweed (Helenium amarum), pale umbrella-sedge (Cyperus acuminatus), marsh bristlegrass (Setaria parviflora), multiflowered false-rhodesgrass

(Trichloris pluriflora), huisache (Acacia farnesiana), Macartney rose (Rosa bracteata), and senna bean (Sesbania drummondii). Species composition varied somewhat by year, especially for annuals and short-lived perennials.

Texas Tech University, Pablo Teveni, August 2017

The site in Refugio County was characterized by a mixture of clay and fine sandy loam surface soils and clay, sandy clay, sandy loam, and sandy clay loam subsurface soils

(USDA 2017). Edroy clay (clay above 178 cm, sandy clay loam between 178 and 203 cm), 0 to 1 percent slopes, ponded, covered 41.8% of the area (9.3 ha; USDA 2017).

Wyick fine sandy loam (fine sandy loam above 25 cm, clay between 25 and 53 cm, sandy clay between 53 and 97 cm, sandy clay loam between 97 and 152 cm) covered 41.3% of the area (9.2 ha; USDA 2017). Banquete clay (clay above 141 cm and fine sandy loam between 141 and 203 cm), 0 to 1 percent slopes, covered 15% percent of the area (3.3 ha;

USDA 2017). Victoria clay (clay above 203 cm), 0 to 1 percent slopes, covered 1.8% of the area (0.4 ha; USDA 2017). These soils all have very slow permeability and medium to high soil available water capacity (Guckian 1988). The ecological sites include

Lakebed PE 25-35”, Claypan Prairie 28-44”, Loamy Prairie 28-40”, and Blackland 24-

44”, and the major land resource area is the Gulf Coast Prairies (USDA 2017). The average annual precipitation at Refugio (24 km SW of study site) is 101.57 cm (U.S.

Climate Data 2017). This site was located in the Southern Subhumid Gulf Coastal

Prairies sub-ecoregion of the Western Gulf Coastal Plain ecoregion (Griffith et al. 2004).

Although the potential climax vegetation is considered to be open grassland dominated by tall, mid, and short grasses, the presence of invasive woody species such as huisache and mesquite indicate continuous heavy grazing in the past (Guckian 1988). The current vegetation can be considered to be a hackberry-huisache woodland association, with nearby ponds and drainage ditches instead of rivers and streams (McLendon 1991). The current vegetation on most of the site includes: huisache/sesbania/shortgrass/sedge community, huisache/sesbania/midgrass community (Reinke 2007b), and

Texas Tech University, Pablo Teveni, August 2017 mesquite/huisache grassland community (Reinke 2007a). The dominant species included: pale umbrella-sedge (Cyperus acuminatus), huisache (Acacia farnesiana), mesquite (Prosopis glandulosa), bermudagrass (Cynodon dactylon), knotgrass (Paspalum distichum), wood-sorrel (Oxalis dillenii), western ragweed (Ambrosia psilostachya), pink evening primrose (Oenothera speciosa), Texas frogfruit (Phyla incisa) eryngo (Eryngium leavenworthii) and senna bean (Sesbania drummondii).

The site in San Patricio County had only clay soils at surface and subsurface:

Victoria clay, 0 to 1 percent slopes, on 85% of the area (12.2 ha) and Monteola clay (clay above 203 cm), 3 to 5 percent slopes, on 15% percent of the area (2.2 ha; USDA 2017).

Although both have very slow permeability, Monteola clay is moderately well-drained and has a medium soil available water capacity while Victoria clay is poorly-drained and has a high soil available water capacity (Guckian and Garcia 1979). The ecological sites include both Blackland 24-44” and Blackland 25-35”, and the major land resource areas include both the Gulf Coast Prairies and the Northern Rio Grande Plain (USDA 2017).

The average annual precipitation at Sinton (16 km SW of study site) is 87.00 cm (U.S.

Climate Data 2017). This site was located in the Southern Subhumid Gulf Coastal

Prairies sub-ecoregion of the Western Gulf Coastal Plain ecoregion (Griffith et al. 2004).

The current vegetation on most of the site includes mesquite/huisache grassland community and mesquite/huisache complex community (Reinke 2007a). Dominant species included: huisache (Acacia farnesiana), mesquite (Prosopis glandulosa), violet ruellia (Ruellia nudiflora), silverleaf nightshade (Solanum elaeagnifolium), woolly croton

(Croton capitatus var. lindheimeri), pink evening primrose (Oenothera speciosa), marsh bristlegrass (Setaria parviflora), buffalograss (Buchloe dactyloides) and upright prairie

Texas Tech University, Pablo Teveni, August 2017 coneflower (Ratibida columnifera).

The site in Kleberg County had sandy loam surface soils and sandy clay loam and clay loam subsurface soils (USDA 2017). Delfina fine sandy loam (fine sandy loam above 41 cm and sandy clay loam between 41 and 203 cm), 0 to 2 percent slopes, covered 44% of the area (33.8 ha; USDA 2017). Colmena fine sandy loam (fine sandy loam above 28 cm and sandy clay loam between 28 and 203 cm), 0 to 1 percent slopes, covered 15% of the area (11.1 ha; USDA 2017). Gertrudis fine sandy loam (fine sandy loam above 43 cm, sandy clay loam between 43 and 103 cm, and clay loam between 103 and 203 cm), 0 to 3 percent slopes, covered 11% of the area (8.5 ha; USDA 2017).

Ecological sites include: Sandy Loam 25-35”, Salty Prairie 26-48”, Gray Sandy Loam

20-35”, Clay Loam 20-25”, and Claypan Prairie 28-44” (USDA 2017). The major land resource areas include: Sandsheet Prairie, Gulf Coastal Prairies, and the Northern Rio

Grande Plain (USDA 2017). The average annual precipitation at Kingsville (19 km SE of study site) is 73.61 cm (U.S. Climate Data 2017). This site was located in the

Southern Subhumid Gulf Coastal Prairies sub-ecoregion in the Western Gulf Coastal

Plain ecoregion (Griffith et al. 2004). The current vegetation on most of the site includes savanna grassland community, short/midgrass shrubland complex, 20-50% canopy, and shrubland complex, > 50% canopy (Carter and Holt undated). Dominant species included: huisache (Acacia farnesiana), Texas crownbeard (Verbesina microptera), marsh bristlegrass (Setaria parviflora), buffelgrass (Cenchrus ciliaris), multiflowered false-rhodesgrass (Trichloris pluriflora), old man’s beard (Clematis drummondii), and

Texas lantana (Lantana horrida).

Texas Tech University, Pablo Teveni, August 2017

Phenology

Every month the average phenological stage of huisache plants between 0.5 m and 2 m in height at each site were determined by recording that of 20 randomly selected plants. Phenological stages included: dormant, bud-break, leaves expanded, flower buds present, flowers expanded, green pods, mature pods, long-shoot stage, short-shoot stage, and senescent. Plants were dormant when no live leaves were present; plants were in bud-break when leaf buds were opening, but leaves had not yet fully expanded. Plants were considered to be in the long shoot stage when vegetative shoots were growing and were considered to be in the short shoot stages when reproductive shoots were growing.

Plants were senescent when defoliation was occurring but live leaves were still present.

Weather Stations

Weather stations (HOBO Micro Station Data Logger H21-002, Onset Computer

Corp., Bourne, MA) were constructed to record precipitation (0.2 mm Rainfall (2m cable)

Smart Sensor S-RGB-M002, Onset Computer Corp., Bourne, MA), air temperature

(Temperature (6’) Sensor TMC6-HE, Onset Computer Corp., Bourne, MA), soil temperature (same sensor type as for air temperature), and volumetric soil water content

(10HS Soil Moisture Smart Sensor S-SMD-M005, Onset Computer Corp., Bourne, MA) at 0.3 m below soil surface. Weather variables were used in the data analysis, along with phenological stages, root carbohydrate content analysis, and herbicide-induced mortality to help determine the most effective environmental conditions for treating huisache with chemicals.

Texas Tech University, Pablo Teveni, August 2017

Carbohydrate Analysis

Every month between April 2012 and November 2014, 5 randomly selected plants

(0.5 m to 2 m in height) from each site, corresponding to the average phenological stage, were excavated manually. A section (15 to 20 cm in length) of the root basal crown was removed using a pick and a hand saw, and root samples were immediately placed in labeled paper sacks and then placed on dry ice in an ice chest to stop enzyme activity.

Frozen samples were transported to Texas Tech University, where they were dried in a forced-air oven at 55 to 60°C (King et al. 2014; Kononoff and Heinrichs 2003) for at least 3 months (Pers. comm. R.E. Sosebee 2012) to halt enzymatic activity and preserve carbohydrates (Wan and Sosebee 1990). The cortex was then separated from the xylem and epidermis using a knife blade or chisel and saved for total nonstructural carbohydrate

(TNC) analysis (Wan and Sosebee 1990) in air-tight plastic containers. Cortex samples were then ground in a Wiley Mill (1-mm screen; Arthur A. Thomas Co., Philadelphia,

PA) (Kononoff and Heinrichs 2003). TNC was extracted from 0.5 g samples through acid hydrolysis by using a reflux apparatus (Lab-line® Multi-Unit Extraction Heater,

Lab-line Instruments, Inc., Melrose Park, IL) to boil each sample in 60 ml. of 0.2N HCl in 300-ml boiling flasks for two hours, as performed by Smith et al. (1964). Samples were cooled and Whatman No. 2 filter paper (General Electric Co., Schenectady, NY) was used to filter the samples into 100-mL volumetric flasks; distilled water was used to increase the volume in each flask to 100 ml and the solution was agitated to ensure homogeneity. An aliquot of 1 ml was then transferred from the flask to 4-ml of distilled water in a 35-ml test tube. The test tube was well-agitated using a vortex mixer

(ThermolyneTM Maxi Mix 1 Type 16700 Mixer, SYBRON Intl. Corp., Milwaukee, WI).

Texas Tech University, Pablo Teveni, August 2017

A 1-ml aliquot was then taken from the first test tube and put into a second test tube containing 10 ml of anthrone reagent, agitated with the vortex mixer, heated at 96 to 100

C in a heating block (Talboys Standard Dry Block Heater, Henry Troemner, LLC,

Thorofare, NJ) for 17 minutes, and cooled to room temperature in a cold water bath. A spectrophotometer (Varian Cary® 50 UV-Visible Spectrophotometer, Agilent

Technologies, Santa Clara, CA) set at 612 nm was then used to measure TNC colorimetrically, using a glucose (dextrose) standard (Murphy 1958). TNC values were reported as percentage of structural dry mass (Hoffman et al. 2003).

Chemical Control Demonstration

Two herbicide formulations that can potentially root-kill huisache were selected with the advice of representatives from Dow AgroSciences (Indianapolis, IN) and

DuPont Crop Protection (Wilmington, DE) chemical companies. The Dow AgroSciences formulation consisted of: GrazonNext HL (431 ml per ha (33.6 oz./acre); active ingredients: aminopyralid + 2,4-D), Tordon 22K (156 ml per ha (13 oz./acre); active ingredient: picloram), and Grazon P+D (862 ml per ha (72 oz./acre); active ingredients: picloram + 2,4-D). The DuPont Crop Protection formulation (currently owned by Bayer

CropScience (Leverkusen, Germany)) consisted of: MAT28 (144 ml per ha (12 oz./acre); active ingredient: aminocyclopyrachlor). Nonionic surfactant was added to all herbicide formulations at 0.5% v/v; blue or purple dye was also added to all herbicide formulations.

Every month ten randomly selected plants per site were sprayed with each herbicide formulation to specification (five plants per site per month were sprayed with the DuPont formulation at Victoria and Refugio Co. sites because exclosure fences had to be built

Texas Tech University, Pablo Teveni, August 2017 and installed, requiring extra time and materials); plants were permanently marked with numbered aluminum tags; and locations were recorded with high-precision GPS receivers

(Magellan eXplorist 310 Handheld GPS Navigator, MiTAC Intl. Corp., Santa Clara, CA).

Individual plants were treated with a hand-held sprayer (Roundup® Premium 2-Gallon

Pump Sprayer, Monsanto Co., St. Louis, MO), all foliage on treated plants was sprayed until wet, and spraying was timed to not take place just prior to precipitation events, though this was not always possible, due to the large amount of travel and specific temporal windows required for this project.

Root mortality of sprayed plants was evaluated two growing seasons (at least 20 months) following herbicide treatment. Resprouting indicated living shrubs. Due to the difficulty of locating previously-treated shrubs (which were often killed or top-killed), treated plants were also marked with rebar, beginning in August 2013 (Pers. comm. C.W.

Hanselka 2013). In order to record complete data sets for analysis, plants that were not found when evaluating mortality were considered to be dead because aerial portions of dead plants were sometimes detached and disintegrated, with the aluminum tags missing

(Pers. comm. C.W. Hanselka 2013).

Analysis

This experiment was analyzed in a randomized complete block design, with site as the blocking factor. This helped control for soil and climatic variability across sites.

Huisache root crown TNC and huisache root mortality data were both subjected to single factor analysis of variance (ANOVA) to test for effect of phenological stage, and

Duncan’s Multiple Range test was used to separate significant (P≤0.05) mean differences

Texas Tech University, Pablo Teveni, August 2017 when significant F-values were found (SPSS 15.0, SPSS Inc., Chicago, IL). Huisache root mortality was also subjected to single-factor ANOVA to test for effect of herbicide formulation (SPSS 15.0, SPSS Inc., Chicago, IL). Root crown TNC and herbicide- induced mortality were both also subjected to single-factor ANOVA to test for effect of month*year, and either Fisher’s Least Significant Difference (LSD) test or Duncan’s

Multiple Range test was used to separate significant mean differences when significant F- values were found (Engle and Bonham 1980; SPSS 15.0, SPSS Inc., Chicago, IL).

Fisher’s LSD test was used to separate means for root-crown TNC because it is a pairwise comparison procedure, and the increase in TNC from one month to the next was the component of interest. Duncan’s Multiple Range test was used to separate means for mortality and for phenological stages because it is a multiple comparison procedure, and differences among all means were the components of interest.

Akaike’s Information Criterion (AIC) was used to determine the model(s) which best fit the mortality data for selected imputs (phenological stage, root crown TNC, weather station data; Excel 2013, Microsoft Corp., Redmond, WA). AIC analysis was performed using multiple data sets with different sample sizes, not only to separate different parameters of interest (e.g. soil type or month) but also to account for missing data from malfunctioning weather station sensors and omitted mortality rates resulting from dates where precipitation influenced herbicide efficacy. The results of these analyses provide for an assessment of the best timing for landowners in the region to most effectively control huisache.

Texas Tech University, Pablo Teveni, August 2017

Study 2: Huisache Seed Germination

This study consisted of cleaned huisache seeds placed in 10 cm x 10 cm plastic petri dishes (Polystyrene container, Hoffman Manufacturing Inc., Jefferson OR), containing blue blotter paper (Hoffman Manufacturing Inc., Jefferson OR) and distilled water, and exposed to diurnally-fluctuating, day/night temperature regimes of 35/25,

30/20, 25/15, and 20/10 degrees Celsius in a growth chamber (SG30SC Controlled

Environment Chamber, Hoffman Manufacturing Inc., Jefferson OR). The seeds were from an unknown provenance in south Texas, meaning that seed characteristics may have differed somewhat from seeds of huisache plant growing in the Coastal Bend region of south Texas. The temperature treatments of 35/25°C , 30/20°C, 25/15°C, and 20/10°C were selected to resemble the high/low temperatures in 1) June, July, and August, 2)

May, September, and early October, 3) late October, November, March, and April and 4)

December, January, and February, respectively (U.S. Climate Data 2017). Temperatures higher than 35°C or lower than 10°C were not used in this project because the growth chamber was not safely usable beyond this temperature range and because average monthly high/low temperatures under field conditions in the Coastal Bend rarely fell outside this range.

Prior to placement in the germination chamber, seeds were either non-pretreated or pretreated by scarification with concentrated sulfuric acid for approximately 25 minutes. One hundred seeds were used for each temperature treatment (20 in each of 5 replicate petri dishes), with all seeds under constant humidity and alternating light and dark photoperiods (12 hours light/12 hours dark). Seeds were checked daily to quantify

Texas Tech University, Pablo Teveni, August 2017 number of seeds germinated and to add enough water so that seeds could respire but were not moisture-limited. The germination of individual seeds was recorded at the first indication of radicle emergence. Percent germination was calculated. ANOVA for a completely randomized design was conducted on total cumulative germination, cumulative germination * days in germination chamber, and non-cumulative germination

* days in germination chamber (SPSS 15.0, SPSS Inc., Chicago, IL). Treatment means were separated using Duncan’s multiple range test at P=0.05 (SPSS 15.0; SPSS Inc.,

Chicago, IL).

Study 3: Huisache Seedling Resprouting and Cotyledon Latent Meristem Positioning

This study began by growing huisache seedlings under greenhouse conditions.

Seeds were scarified in sulfuric acid for 25 minutes, rinsed thoroughly, and sown 2 cm deep (Scifres 1974) in deep containers (35.6 cm depth x 6.9 cm dia.; Large Deepot Cells

D60H, Stuewe and Sons, Inc., Tangent, OR) to accommodate the long tap roots of huisache. Plants were fully-irrigated with misters for 15 minutes daily, increasing to 20 minutes daily after 7 months, 25 minutes daily after 14 months and 30 minutes daily after

20 months. A total of 240 seedlings were used to evaluate the resprouting of huisache following the removal of above-ground portions. Once a month, ten seedlings were measured in height, and above-ground portions were clipped at ground-level and placed in labeled paper sacks atop ice in an ice chest for transport to the laboratory. In the laboratory, fresh mass of above-ground portions was measured using a laboratory scale; plants were then dried in a forced-air drying room (55 to 60°C) for at least three months before dry mass was measured using the same scale. Plants were observed for at least six

Texas Tech University, Pablo Teveni, August 2017 months after cutting to determine whether they had the ability to re-sprout from belowground portions.

Another 120 seedlings were grown under similar conditions as in the seedling resprouting study described above to evaluate the burying of the cotyledon latent meristem. Cotyledon levels were marked with permanent marker on all seedlings upon emergence. Every two months, ten plants were measured in height and harvested in total to determine position, relative to the soil surface, of the cotyledon latent meristems. Both fresh and dry mass of plants were measured as described for seedlings in the resprouting study described above.

Analysis

A single factor, completely-randomized design ANOVA was used to test effect of age (in months) on height of cotyledon latent meristems, and where significant, Duncan’s multiple range test was used to separate treatment means at P = 0.05 (SPSS 15.0, SPSS

Inc., Chicago, IL). Probit analysis was used to characterize monthly changes in percent seedlings resprouting and percent of seedlings with cotyledon latent meristems at or below soil level (SPSS 15.0, SPSS Inc., Chicago, IL). Regression analysis was used to test the effect of both height and above-ground dry mass on seedling resprouting, cotyledon latent meristem height, and percent of seedlings with cotyledon latent meristems at soil level (SPSS 15.0, SPSS Inc., Chicago, IL).

Texas Tech University, Pablo Teveni, August 2017

Chapter 3

RESULTS

Study 1: Effective Herbicide Control of Huisache

At all study sites, the most common average phenological stage was “leaves expanded, full canopy” (usually including continuous shoot growth and leaf expansion).

Flowering huisache plants less than 2 m high (and plants with floral buds) were sometimes present, particularly at the Victoria Co. study site; however, those plants were never numerous enough to comprise the average phenological stage at any site for any month. Therefore, because the average phenological stage of huisache plants never included floral buds, flowers, pods, or short shoots (reproductive shoots), the only average phenological stages recorded in this study were: dormant, bud-break, full canopy

(leaves expanded), and senescent. The full canopy phenological stage is equivalent to the long shoot stage. The phenological stage “senescent” might best be described as “partial defoliation” in the case of huisache, because late autumn and winter cold temperatures that cause senescence in temperate deciduous trees cause huisache to lose some or most of its leaves, but shoots and leaves often grow through the cool season or plants retain surviving leaves, without transitioning into true dormancy (personal observation). This continuous growth habit likely results from the plant’s tropical origins. However, plants did appear to become dormant when temperatures were sufficiently cold.

Carbohydrate Analysis and Chemical Control Demonstration – ANOVA Analysis

Total nonstructural carbohydrate (TNC) concentration in huisache root crowns was significantly different between soil types, with TNC in clay significantly higher than

Texas Tech University, Pablo Teveni, August 2017 in sandy loam / sandy clay loam (date not shown). TNC concentration was also significantly different among phenological stages (F(3,633) = 4.375, P = 0.005; Table 1), with TNC during partial defoliation (M = 5.025, SD = 0.9153) and dormancy (M = 5.124,

SD = 0.8997) significantly greater than TNC during bud-break (M = 4.491, SD =

0.9147); TNC during full canopy was intermediate (M = 4.787, SD = 1.032; Figure 1).

Herbicide-induced root mortality in huisache was significantly different between soil types, with mortality of plants growing in predominantly clay soil greater than that of plants growing in predominantly sandy loam / sandy clay loam soil (data not shown).

Herbicide-induced huisache root mortality was significantly different between herbicides

(F(1,180) = 5.177, P = 0.024; Table 2), with mortality of plants sprayed with MAT28 (M

= 83.8%, SD = 24.87) greater than mortality of plants sprayed with the DOW formulation

(M = 74.1%, SD = 31.72; Figure 2A). Mortality was also significantly different among phenological stages (F(3,178) = 23, P < 0.001; Table 2), with greatest mortality rate at

Full Canopy stage (M = 86.1%, SD = 21.19), lowest mortality rate at Dormant stage (M =

34.3%, SD = 26.99), and intermediate mortality at Partial Defoliation (M = 60.6%, SD =

37.88) and Bud-Break (43.3%, SD = 34.45) phenological stages (Figure 2B).

TNC concentration across all sites was significantly different between months

Table 1. Single-factor analysis of variance table for dependent variable Root Crown TNC, across all sites, with fixed factors Phenological Stage and Site (blocking factor). Source of Degrees Type III Mean F-Statistic Significance Variation of Sum of Square Freedom Squares Phenological Stage Phenology 3 12.720 4.240 4.379 0.005 Site 3 26.671 8.890 9.182 < 0.001 Error 633 612.917 0.968

Texas Tech University, Pablo Teveni, August 2017

6 Bud-Break Full Canopy Partial Defoliation Dormant a a 5 b ab

3 (%)

1 Total Nonstructural Carbohydrate NonstructuralTotal 0 Phenological Stage

Figure 1. Average total nonstructural carbohydrate (TNC, %) for different phenological stages. Statistical significance determined using ANOVA with randomized complete block design and Duncan’s multiple range test where appropriate. Bars represent ± SE. * Means with same letter are not significantly different at P = 0.05.

(F(31,605) = 5.023, P < 0.001; Table 3). Fisher’s Protected LSD test revealed significant increases in TNC between the months of April and May in 2012 (M = 0.78%, SE =

0.2868, P = 0.007) and 2014 (M = 0.66%, SE = 0.2868, P = 0.022), between July and

August in 2012 (M = 0.64%, SE = 0.2868, P = 0.027), 2013 (M = 1.37%, SE = 0.2868, P

< 0.001) and 2014 (M = 1.15%, SE = 0.2868, P = 0.012), between November and

December in 2012 (M = 0.73%, SE = 0.2868, P = 0.012), and between March and April in 2013 (M = 0.78%, SE = 0.2868, P = 0.007; Figure 3).

Herbicide-induced huisache root mortality across all sites was also significantly different between months (F(28,153) = 5.748, P < 0.001; Table 4). Duncan’s Multiple

Range test revealed that the greatest mortality rates occurred from September 2012 to

January 2013, from April to December in 2013, from April to July in 2014, and from

September to November in 2014; within these time periods mortality was especially high

Texas Tech University, Pablo Teveni, August 2017

Table 2. Single-factor analysis of variance table for dependent variable Huisache Mortality, across all sites, with fixed factors Herbicide Formulation, Phenological Stage, and Site (blocking factor).

Source of Degrees Type III Mean F-Statistic Significance Variation of Sum of Square Freedom Squares Herbicide Formulation Month 1 0.422 0.422 5.177 0.024 Site 3 0.936 0.312 3.822 0.011 Error 180 14.688 0.082 Phenological Stage Phenology 3 4.221 1.407 23 < 0.001 Site 3 1.125 0.375 6.13 0.001 Error 178 10.889 0.061

in October 2013 (M = 100%, SD = 0), October 2014 (M = 100%, SD = 0), and May 2014

(M = 96.25%, SD = 7.44; Figure 3).

Root crown TNC analysis of huisache growing in sandy loam / sandy clay loam sites was significantly different between months (F(31,287) = 4.353, P < 0.001; Table 5).

100 DOW DuPont a a Bud-Break * 80 b Full Canopy Partial Defoliation b Dormant 60 bc c 40

20 Huisache Root Mortality (%) Mortality Root Huisache A B 0 Herbicide Phenological Stage

Figure 2. Average total herbicide-induced huisache mortality (%) for A) Herbicide Formulation, and B) Phenological Stage. Statistical significance determined using ANOVA with randomized complete block design and Duncan’s multiple range test where appropriate. Bars represent ± SE. * Means with same letter are not significantly different at P = 0.05.

Texas Tech University, Pablo Teveni, August 2017

Table 3. Single-factor analysis of variance table for dependent variable Huisache Root Crown TNC, across all sites, with fixed factors Month*Year and Site (blocking factor). Source of Degrees of Type III Mean F-Statistic Significance Variation Freedom Sum of Sums of Squares Squares Month*Year 31 128.063 4.131 5.023 < 0.001 Site 3 24.956 8.319 10.115 < 0.001 Error 605 497.574 0.822

Fisher’s Protected LSD test revealed significant increases in TNC between the months of

April and May 2012 (M = 1.56%, SE = 0.3996, P < 0.001), July and August in 2012 (M

= 0.81%, SE = 0.3996, P = 0.044), 2013 (M = 1.13%, SE = 0.3996, P = 0.005), and 2014

(M = 1.31%, SE = 0.3996, P = 0.001), November and December 2012 (M = 0.98%, SE =

0.3996, P = 0.015), and March and April 2013 (M = 1.1%, SE = 0.3996, P = 0.006;

Figure 4).

Herbicide-induced mortality of huisache growing in sandy loam / sandy clay loam sites was significantly different between months (F(28,64) = 4.879, P < 0.001; Table 6).

According to Duncan’s Multiple Range test, greatest mortality rates occurred from July

2012 to January 2013, April to December 2013, April to July 2014, and September to

November 2014, with the highest mortality rates in October 2014 (M = 100%, SD = 0),

June 2014 (M = 100%, SD = 0), October 2013 (M = 100%, SD = 0), and May 2014 (M =

Table 4. Single-factor analysis of variance table for dependent variable Huisache Mortality, across all sites, with fixed factors Month*Year and Site (blocking factor). Source of Degrees Type III Mean Sums F-Statistic Significance Variation of Sum of of Squares Freedom Squares Month*Year 28 7.746 0.277 5.748 < 0.001 Site 3 0.817 0.272 5.659 0.001 Error 153 7.364 0.048

Texas Tech University, Pablo Teveni, August 2017

Table 5. Single-factor analysis of Variance table for dependent variable Huisache Root Crown TNC, across both sites with predominantly sandy loam / sandy clay loam soil (Diebel Ranch and King Ranch), with fixed factors Month*Year and Site (blocking factor). Source of Degrees of Type III Mean F-Statistic Significance Variation Freedom Sum of Sums of Squares Squares Month*Year 31 107.715 3.475 4.353 <0.001 Site 1 6.641 6.641 8.32 0.004 Error 287 229.096 0.798

100%, SD = 0; Figure 4).

Herbicide-induced mortality of huisache growing in sandy loam / sandy clay loam sites was significantly different between months (F(28,64) = 4.879, P < 0.001; Table 6).

According to Duncan’s Multiple Range test, greatest mortality rates occurred from July

2012 to January 2013, April to December 2013, April to July 2014, and September to

November 2014, with the highest mortality rates in October 2014 (M = 100%, SD = 0),

June 2014 (M = 100%, SD = 0), October 2013 (M = 100%, SD = 0), and May 2014 (M =

100%, SD = 0; Figure 4).

Root crown TNC concentrations of huisache growing in clay sites were significantly different between months (F(31,287) = 3.736, P < 0.001; Table 7).

According to Fisher’s Protected LSD, TNC concentrations increased significantly

Table 6. Single-factor analysis of variance table for dependent variable Huisache Mortality, across both sites with predominantly sandy loam / sandy clay loam soil (Diebel Ranch and King Ranch), with fixed factors Month*Year and Site (blocking factor). Source of Degrees of Type III Mean F-Statistic Significance Variation Freedom Sum of Square Squares Month*Year 28 6.71 0.24 4.879 <0.001 Site 1 0.391 0.391 7.966 0.006 Error 64 3.144 0.049

Texas Tech University, Pablo Teveni, August 2017

Table 7. Single factor analysis of variance table for dependent variable Huisache Root Crown TNC, across both sites with predominantly clay soil (Baumgartner Ranch and Welder Refuge), with fixed factors Month*Year and Site (blocking factor). Source of Degrees of Type III Mean F-Statistic Significance Variation Freedom Sum of Square Squares Month*Year 31 83.038 2.679 3.736 < 0.001 Site 1 1.77 1.77 2.469 0.117 Error 287 205.788 0.717

between the months of May and June 2012 (M = 0.9%, SE = 0.3787, P = 0.018), July and

August in 2013 (M= 1.61%, SE = 0.3787, P < 0.001) and in 2014 (M =1.04%, SE =

0.3787, P = 0.006), September and October 2013 (M = 0.75%, SE = 0.3787, P = 0.049), and April to May 2014 (M = 0.91%, SE = 0.3787, P = 0.017; Figure 5).

Herbicide-induced mortality of huisache growing in clay sites was significantly different between months (F (27,62) = 2.46, P = 0.002; Table 8). Mortality was greatest in July 2012, October 2012 to December 2013, and April to November 2014, especially in June (M = 97.5%, SD = 5.0) and November (97.5%, SD = 5.0) 2013 and October 2012

(M = 95%, SD = 7.071; Figure 5).

Weather data from all sites showed seasonal fluctuations in soil temperature and precipitation (Figure 6). The results of analyses of the effect of weather variables on herbicide-induced mortality in huisache were included in the next section.

Table 8. Single-factor analysis of variance table for dependent variable Huisache Mortality, across both sites with predominantly clay soil (Baumgartner Ranch and Welder Refuge), with fixed factors Month*Year and Site (blocking factor). Source of Degrees of Type III Mean F-Statistic Significance Variation Freedom Sum of Square Squares Month*Year 27 2.719 0.101 2.46 0.002 Site 1 0.152 0.152 3.716 0.058 Error 62 2.538 0.041

Texas Tech University, Pablo Teveni, August 2017

6.5 100 TNC Mortality bcdef◊◊ abcd 2012 6 ◊ cdef abc 80 5.5 ** abcd abcd 5 60 * 4.5 40 4 * 20 3.5 ←------[2]Ŧ------→[2/3] 3 0

6.5 100 abcde a abcd 2013 6 abcde fgh defg ab abc ab 80 5.5

5 60 ab 4.5 ** *** abcd abc 40

Carbohydrate (%) Carbohydrate Mortality Rate (%) Rate Mortality

Total Nonstructural Total 20 3.5 ←---[3]---→ [1] ←------[2]------→ [3] 3 0

6.5 100 abcd 2014 6 a ab bcdef abc 80 5.5 abc gh efgh abcd 5 abcd 60 h * *** 4.5 40 4 20 3.5 [3/4] [3] [1] ←------[2]------→ [3] 3 0

Month

Figure 3. Average monthly total nonstructural carbohydrate content (TNC, %) and herbicide-induced mortality (%) across all study sites. Statistical significance was determined using ANOVA with randomized complete block design and Fisher’s Protected LSD test for TNC and Duncan’s Multiple Range test for mortality. Bars represent ± SE. ◊ TNC concentration of month is significantly greater than TNC of previous month at *) α = 0.05, **) α = 0.01, or ***) α < 0.001. ◊◊ Mortality rates of months with same letter are not statistically different at α = 0.05. Ŧ Phenology: [1] = bud-break, [2] = leaf expansion (full canopy), [3] = partial defoliation, [4] = dormant

Texas Tech University, Pablo Teveni, August 2017

6.5 100 ****◊ abc 2012 TNC Mortality 80 5.5 ab 60 4.5 abcd 40 abcd 3.5 * * abcd◊◊ abcd 20 ←------[2]Ŧ------→ [3] 2.5 0 6.5 ** 100 2013 a abcd a abcd cde ** 80 5.5 ab abc 60 4.5 abcd 40 abcd e abc (%) Mortality Carbohydrate (%) Carbohydrate 3.5 Total Nonstructural Total 20 abcd [3/4] [3] [1] ←------[2]------→ [3] 2.5 0

6.5 100 2014 [3/4] a abc a ab 80 5.5 *** ab ab de 60 4.5 40 e abcd 3.5 bcde e 20 [3] 2.5 [3] [1] ←------[2]------→ 0

Month Figure 4. Average monthly total nonstructural carbohydrate content (TNC, %) and herbicide-induced mortality (%) for predominantly sandy loam / sandy clay loam sites. Statistical significance determined using ANOVA with randomized complete block design and Fisher’s Protected LSD test for TNC and Duncan’s Multiple Range test for mortality. Bars represent ± SE. ◊ TNC concentration of month is significantly greater than TNC of previous month at *) α = 0.05, **) α = 0.01, ***) or α = 0.001, ****) α < 0.001. ◊◊ Mortality rates of months with same letter are not statistically different at α = 0.05. Ŧ Phenology: [1] = bud-break, [2] = leaf expansion (full canopy), [3] = partial defoliation, [4] = dormant

Texas Tech University, Pablo Teveni, August 2017

6.5 100 2012 a ◊ d 80 5.5 * ab 60 abc 4.5 abcd◊◊ 40 3.5 20 ←------[2]Ŧ------→[2/3] 2.5 0 6.5 100 2013 a a a a abcd 80 5.5 a * ab ab 60 4.5 abcd abc 40

*** Mortality (%) Mortality

Carbohydrate (%) Carbohydrate 3.5

Totatl Nonstructural Totatl a abcd 20 ←--[3]--→ [1] ←------[2]------→ [3] 2.5 0

6.5 100 a abc a 2014 abcd * 80 5.5 bcd cd ab ab abc ** ab 60 4.5 40 3.5 20 ←[3/4]→ [1] ←------[2]------→ [3] 2.5 0

Month

Figure 5. Average monthly total nonstructural carbohydrate content (TNC, %) and herbicide-induced mortality (%) for predominantly clay sites. Statistical significance determined using ANOVA with randomized complete block design and Fisher’s Protected LSD test for TNC and Duncan’s Multiple Range test for mortality. Bars represent ± SE. ◊ TNC concentration of month is significantly greater than TNC of previous month at *) α = 0.05, **) α = 0.01, or ***) α < 0.001. ◊◊ Mortality rates of months with same letter are not statistically different at α = 0.05. Ŧ Phenology: [1] = bud-break, [2] = leaf expansion (full canopy), [3] = partial defoliation, [4] = dormant

Texas Tech University, Pablo Teveni, August 2017

40 120

100 C)

° 30 80 20 60 40

10 Temperature ( Temperature

20 (mm) Perecipitation A 0 0

40 150 C) ° 30 100 20 50

Temperature ( Temperature Precipitation (mm) Precipitation B 0 0

40 150 C)

° 30 100 20 50

Temperature ( Temperature Precipitation (mm) Precipitation C 0 0 40 90

80 C) ° 30 70 60 50 20 40 30

10 20 Temperature ( Temperature

10 (mm) Precipitation D 0 0

Date Figure 6. Average daily soil temperature below 0.3 m and daily precipitation for all four study sites, during period of field study (mid-April 2012 to mid-November 2014): A= Baumgartner Ranch, B = Diebel Ranch, C = King Ranch, and D = Welder Wildlife Refuge.

Texas Tech University, Pablo Teveni, August 2017

Carbohydrate Analysis and Chemical Control Demonstration – AIC Analysis

Best-fit AIC results for herbicide-induced mortality are presented for all sites

(Figures 7 and 8), predominantly sandy loam sites (Figures 9 and 10), predominantly clay sites (Figure 11), all plants in full canopy phenological stage (Figure 12), all plants in partially-defoliated phenological stage (Figure 13), and from all sites treated in the months of high herbicide mortality (Figures A-2 to A-5). Some datasets had more than one model because 1) the parameter “month” can vary greatly from year to year in regard to weather and other factors and is therefore not as stable as predictor as is for example, soil temperature (hence, a second model was also chosen if the first model included month as a parameter [i.e. Figures 7, 8, 9 and 10]), and 2) some parameters were not comparable to each other due to missing data, especially from malfunctioning weather sensors; therefore, because it could be impossible in such cases to determine which model was better (due to different sample sizes) the results of both models were presented (i.e.

Figures A-2 to A-5)

The best model fit for chemically-induced huisache mortality across all dates and sites was a sixth-order polynomial of mortality vs. month (Figure 7), with AIC =

1169.382, AICc (AIC corrected for small sample sizes) = 1170.015, R2 = 0.4054, Adj. R2

= 0.3854, and n = 185. The equation for the function is:

y = 0.006x6 – 0.2741x5 + 4.8114x4 – 40.86x3 + 170.69x2 – 305.69x + 214.43

[where y = mortality (%) and x = month (i.e. 1 = Jan., 2 = Feb, 3 = Mar., etc.)]

Overall herbicide-induced huisache mortality was greatest during the months of October,

Texas Tech University, Pablo Teveni, August 2017

November, and May. When month was removed as a model parameter, the best fit model for huisache herbicide mortality across all dates and sites was a combination of two quadratic functions: mortality vs. soil temperature and mortality vs. phenology (Figure

8), with AIC = 1117.623, AICc = 1117.983, R2 = 0.3123, Adj. R2 = 0.301, and n =173.

The equations for the two functions are, respectively:

y = -0.3839x2 + 18.806x – 138.84

[where y = mortality (%) and x = soil temperature (°C) at 0.3 m], and

y = -18.401x2 + 77.854x + 1.8649

[where y = mortality (%) and x = phenological stage (i.e. 1 = bud-break, 2 = full canopy,

3 = partial defoliation, and 4 = dormant)]

When month was removed as a parameter, huisache herbicide mortality was greatest when soil temperature at 0.3 m was 24.49°C and plants were in the leaves expanded (full canopy) phenological stage.

When only the two sites with predominantly sandy loam / sandy clay loam soil

(Diebel Ranch and King Ranch) were analyzed with AIC, the best fit model for herbicide-induced huisache mortality was a six-order polynomial of mortality vs. month, combined with a quadratic function of mortality vs. phenology and a linear function of mortality vs. delta-TNC (current month root crown TNC average – previous month root crown TNC average; Figure 9), with AIC = 594.377, AICc = 597.028, R2 = 0.537, Adj.

R2 = 0.522, and n = 94. The equations for the three functions are, respectively:

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality

0 1 2 3 4 5 6 7 8 9 10 11 12 Month*

Figure 7. Best-Fit Model (Huisache Mortality vs. Month) for all data, according to AIC (Akaike’s Information Criterion) analysis (AIC = 1169.382, AICc [AIC corrected for small sample sizes] = 1170.015, R2 = 0.4054, Adj. R2 = 0.3854, and n = 185). Function: y = 0.006x6 – 0.2741x5 + 4.8114x4 – 40.86x3 + 170.69x2 – 305.69x + 214.43, R2 = 0.4054 * Month: 1 = January, 2 = February, 3 = March, etc.

y = 0.0045x6 – 0.2241x5 + 4.232x4 – 38.195x3 + 167.49x2 – 307.49x + 200.7

[where y = mortality (%) and x = month (i.e. 1 = Jan., 2 = Feb., 3 = Mar., etc.)], and

y = -26.483x2 + 112.95x - 38.587

[where y = mortality (%) and x = phenological stage (i.e. 1 = bud-break, 2 = full canopy,,

3 = partial defoliation, and 4 = dormant], and

y = -1.8323x +73.251

[where y = mortality (%) and x = Delta-TNC (change in average TNC (%) from one

month to the following month)]

When huisache was grown in sandy loam / sandy clay loam soils, herbicide-induced mortality was greatest during the months of November, October, and May, when plants

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality

0 10 15 20 25 30 35 A Soil Temperature (°C) at 0.3 m depth

100

40 Mortality (%) Mortality

0 1 2 3 4 B PhenologyƗ

Figure 8. Best fit model for all data (A) Huisache Mortality vs. Soil Temperature at 0.3 m and B) Huisache Mortality vs. Phenology), according to AIC analysis (AIC = 1117.623, AICc = 1117.983, R2 = 0.3123, Adj. R2 = 0.301, and n =173), when the parameter month is removed from analysis A) Function: y = -0.3839x2 + 18.806x – 138.84, R2 = 0.271 B) Function: y = -18.401x2 + 77.854x +1.8649, R2 = 0.2154 Ɨ Phenology: 1 = budbreak, 2 = leaves expanded (full canopy), 3 = partial defoliation, 4 = dormant

were in leaves expanded (full canopy) phenological stage, and when root crown TNC was lower than in the previous month. When month was removed as a parameter, the best fit model became a quadratic function of mortality vs. soil temperature and a linear function

Texas Tech University, Pablo Teveni, August 2017 of mortality vs. 8-week cumulative precipitation (Figure 10), with AIC = 490.98, AICc =

491.559, R2 = 0.339, Adj. R2 = 0.31, and n = 74. The equations for the two functions are, respectively:

y = -0.425x2 + 21.436x – 181.53

[where y = mortality (%) and x = soil temperature (°C) at 0.3 m], and

y = 0.2005x + 51.316

[where y = mortality (%) and x = cumulative 8 week precipitation (mm)]

When month was removed as a parameter, huisache growing in sandy loam / sandy clay loam soils, had greatest mortality when soil temperature at 0.3 m depth was 25.22° C and cumulative rainfall during the previous 8 weeks was high.

When only the two sites with predominantly clay soil (Baumgartner Ranch and

Welder Refuge) were analyzed with AIC, the best fit model for huisache mortality was a combination of a quadratic function of mortality vs. soil temperature and a linear function of mortality vs. Delta-TNC (Figure 11), with AIC = 511.725, AICc = 512.238, R2 =

0.308, Adj. R2 = 0.282, and n = 83. The equations are, respectively:

y = -0.3946x2 + 18.595x - 122.79, R2 = 0.277

[where y = mortality (%) and x = soil temperature (°C) at 0.3 m], and

y = 5.9117x + 82.09, R2 = 0.0452

[where y = mortality (%) and x = Delta-TNC (change in average TNC (%) from one

month to the following month)]

Texas Tech University, Pablo Teveni, August 2017

100

Mortality (%) Mortality 20

0 A 1 2 3 4 5 6 7 8 9 10 11 12 MonthƗ 100

Mortality (%) Mortality 20

0 1 2 3 4 B ƗƗ Phenology 100

Mortality (%) Mortality 20

0 C -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 Delta-TNC (%) Figure 9. Best fit model for data from predominantly sandy loam / sandy clay loam sites (A) Huisache Mortality vs. Month, B) Huisache Mortality vs. Phenology, and C) Huisache Mortality vs. Delta-TNC [current month TNC – previous month TNC), according to AIC analysis (AIC = 594.377, AICc = 597.028, R2 = 0.537, Adj. R2 = 0.522, and n = 94) A) Function: y = 0.0045x6 – 0.2241x5 + 4.232x4 – 38.195x3 + 167.49x2 – 307.49x + 200.7, R2 = 0.537 B) Function: y = -26.483x2 + 112.95x – 38.587, R2 = 0.2987 C) Function: y = -1.8323x + 73.251, R2 = 0.0027 Ɨ Month: 1 = January, 2 = February, 3 = March, etc. ƗƗ Phenology: 1 = budbreak, 2 = leaves expanded (full canopy), 3 = partial defoliation, 4 = dormant

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality

0 10 15 20 25 30 35 A Soil Temperature (°C) at 0.3 m

100

40 Mortality (%) Mortality

0 0 50 100 150 200 250 300 B 8-week Cumulative Precipitation (mm)

Figure 10. Best fit model for data from predominantly sandy loam / sandy clay loam sites (A) Huisache Mortality vs. Soil Temperature at 0.3 m and B) Huisache Mortality vs. Cumulative 8-week Precipitation), according to AIC analysis (AIC = 490.98, AICc = 491.559, R2 = 0.339, Adj. R2 = 0.31, and n = 74), when the parameter month is removed from analysis A) Function: y = -0.425x2 + 21.436x – 181.53, R2 = 0.3047 B) Function: y = 0.2005x + 51.316, R2 = 0.155

When growing in predominantly clay soils, herbicide-induced huisache mortality was greatest when soil temperature at 0.3 m is 23.85° C and root crown TNC was higher than in the previous month. 53

Texas Tech University, Pablo Teveni, August 2017

When only data from the full canopy phenological stage were analyzed with AIC, the highest fit model was a sixth-order polynomial function of mortality vs. month, with

AIC = 819.776, AICc = 820.645, R2 = 0.1958, Adj. R2 = 0.1587, and n = 137 (function and figure not shown). This model was similar to the best-fit model for all data (Figure

7), except greatest herbicide-induced mortality was during the months of December,

October, May, and November (instead of only October, November, and May). Because this sixth-order polynomial function may actually over-fit the data; a fifth-order polynomial (function and figure not shown) may be a more appropriate model for the data, (similar to best fit model for all data [Figure 7], except greatest mortality was during months of November, October, December, and May [instead of October, November, and

May]), despite having greater AIC values. According to AIC analysis, the fifth-order polynomial did not fit the data as well as the following model. When month was removed as a parameter, the best fit model according to AIC analysis was a quadratic function of mortality vs. soil temperature (Figure 12), with AIC = 749.722, AICc =

749.92, R2 = 0.1023, Adj. R2 = 0.0875, and n = 125. The function for this model is:

y = -0.3491x2 +16.519x – 102.42

[where y = mortality (%) and y = soil temperature (°C) at 0.3 m]

When huisache were in the full canopy phenological stage, herbicide-induced mortality was greatest when soil temperature at 0.3 m was 23.66 °C.

Texas Tech University, Pablo Teveni, August 2017

100 80 60 40

Mortality (%) Mortality 20 0 10 15 20 25 30 35 A Soil Temperature (°C) at 0.3 m

100 80 60 40

Mortality (%) Mortality 20 0 -3 -2 -1 0 1 2 3 B Delta-TNC (%)

Figure 11. Best fit model for data from predominantly clay sites (A) Huisache Mortality vs. Soil Temperature at 0.3 m and B) Huisache Mortality vs. Delta-TNC), according to AIC analysis (AIC = 511.725, AICc = 512.238, R2 = 0.308, Adj. R2 = 0.282, and n = 83) A) Function: y = -0.3946x2 + 18.595 – 122.79, R2 = 0.277 B) Function: y = 5.9117x + 82.09, R2 = 0.0452

When only data from plants in the partially-defoliated phenological stage were analyzed with AIC, the best fit model was a combination of a linear function of mortality vs Future Delta-TNC (TNC of next month – TNC of current month) and a linear function of mortality vs. 4-week cumulative precipitation (Figure 13), with AIC = 170.659, AICc

= 171.859, R2 = 0.3529, Adj. R2 = 0.2934, and n = 24. These linear equations are, respectively:

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality 20

0 10 15 20 25 30 35 Soil Temperature (°C) at 0.3 m

Figure 12. Best fit model for full canopy phenological stage data from all sites (Huisache Mortality vs. Soil Temperature), according to AIC analysis (AIC = 749.722, AICc = 749.92, R2 = 0.1023, Adj. R2 = 0.0875, and n = 125). Function: -0.3491x2 + 16.519x – 102.42, R2 = 0.1023

y = 22.923x + 58.179

[where y = mortality (%) and x = Future Delta-TNC (change in TNC(%) in one month

from the previous month)], and

y = 0.7883x + 26.669

[y = mortality (%) and x = cumulative 4-week precipitation (mm)]

When huisache was in the partial defoliation phenological stage, herbicide-induced mortality was greatest when average root crown TNC was lower in the current month than in the following month and when cumulative 4-week precipitation was high.

When data from plants treated in May, October, and November were each analyzed separately with AIC, mortality models differed in the parameters that were most important in the model. The best-fit models for these high-mortality months of treatment

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality 20

0 -2 -1.5 -1 -0.5 0 0.5 1 A Future Delta-TNC (%)

100

40 Mortatliy (%) Mortatliy 20

0 0 10 20 30 40 50 60 70 80 B 4-Week Cumulative Precipitation (mm)

Figure 13. Best fit model for data from plants in partial defoliation phenological stage from all sites (A) Huisache Mortality vs. Future-Delta-TNC [next month TNC – current TNC] and B) Huisache Mortality vs. 4-week Cumulative Precipitation), according to AIC analysis (AIC = 170.659, AICc = 171.859, R2 = 0.3549, Adj. R2 = 0.2934, and n = 24) A) Function: y = 22.923x +58.179, R2 = 0.2165 B) Function: y = 0.7883x + 26.669, R2 = 0.1587

included: a negative linear function of mortality vs. Future-Delta-TNC, a negative linear function of mortality vs. 4-week cumulative precipitation, a positive linear function of mortality vs. soil available water content (%; AWC) at 0.3 m depth, and a positive linear function of mortality vs. soil temperature at 0.3 m (Figures A-2 to A-5).

Texas Tech University, Pablo Teveni, August 2017

Study 2: Huisache Seed Germination

Cumulative total germination rates were significantly different between scarified and non-scarified treatments and among temperature*scarified treatments (F(7,32) =

185.722, P < 0.001; Table 9), but were not significantly different among temperature treatments. Cumulative total germination rates for scarified seeds were: 96% (SD =

5.4772) for 35/25, 96% (SD = 6.5192) for 30/20, 99% (SD = 2.2361) for 25/15, and 97%

(SD = 4.4721) for 20/10°, while cumulative total germination rates for non-scarified seeds were: 19% (SD = 9.6177) for 35/25, 21% (SD = 8.9443) for 30/20, 20% (SD = 7.9057) for 25/15, and 16% (SD = 6.5192) for 20/10 (Figure 14).

Although no differences in cumulative total germination were evident between temperature treatments, germination curves revealed temporal differences in rate of germination between temperature treatments (Figure 15). Significant differences in cumulative germination were found when temperature treatment * days in growth chamber were analyzed for scarified seeds (F(167,672) = 169.601 , P < 0.001), and for non-scarified seeds (F(167,672) = 4.809, P < 0.001, Table 10). However, when non- cumulative germination percentages were analyzed for temperature treatment*days in growth chamber, significant differences were still found for scarified seeds (F(167,672) =

Table 9. Single-factor analysis of variance table for dependent variable Cumulative Total Germination, with fixed factors Scarification * Temperature. Source of Degrees Type III Mean F-Statistic Significance Variation of Sum of Square Freedom Squares Scarification * Temperature Between Groups 7 60940 8705.714 185.722 < 0.001 Within Groups 32 1500 46.875

Texas Tech University, Pablo Teveni, August 2017

52.644, P < 0.001) but not for non-scarified seeds (F(167,672) = 1.106, P = 0.196; Table

10). On day 3, 35/25-Scarified had a significantly greater germination percentage than all other scarified treatments (M = 25, SD = 10.6066). 25/15 Scarified had significantly higher germination percentages than 30/20 Scarified and 20/10 Scarified from day 3 to day 22. On days 3 to 11, 35/25 Scarified and 25/15 Scarified both had significantly greater germination percentages than the other two scarified treatments. 20/10 Scarified had significantly lower germination percentages than the other scarified treatments from day 4 until day 16. Following day 25, scarified temperature treatments were no longer significantly different from each other (Table 11). Germination percentages for non-

100 a◊ a a a

60 Scarified NonScarified 40

b b b b

Total Germination (%) Germination Total 20

0 35/25 30/20 25/15 20/10 Temperature Treatment (High/Low,°C)

Figure 14. Cumulative total germination percentages for Scarified and Non-Scarified huisache seeds growing in a germination chamber, under four different temperature regimes (high/low, °C). Statistical significance determined using ANOVA with completely randomized design and Duncan’s Multiple Range test. Bars represent ± SE. ◊ Columns with the same letter are not significantly different at P = 0.05. scarified temperature treatments were only significantly different between different number of days in germination chamber, not between temperature treatments (data not shown). 59

Texas Tech University, Pablo Teveni, August 2017

Study 3: Huisache Seedling Resprouting and Cotyledon Latent Meristem Positioning

Probit analysis average marginal means indicated that month significantly predicted huisache resprouting (M = 4.304, SE = 0.717, 95% CI [2.898, 5.71], z-score =

5.999, P < 0.001). According to the probit model, the percentage of seedlings resprouting following top removal was 1% at 5.8 months (95% CI [3.521, 7.631]), 10% at 10.2 months (95% CI [7.923, 11.855]), and 50% at 20.3 months (95% CI [18.059,

24.145]; Table 12).

Table 10. Single-factor analysis of variance table for dependent variables Cumulative Germination and Non-cumulative Germination, with fixed factors Temperature * No. of Days (in germination chamber), for Scarified and Non-scarified seeds. Source of Degrees Type III Mean F-Statistic Significance Variation of Sum of Square Freedom Squares Cumulative Germination - Scarified) Temperature * No. of Days Between Groups 167 763719.9 4573.173 169.601 < 0.001 Within Groups 672 18120.00 26.964 Cumulative Germination - Non-scarified) Temperature * No. of Days Between Groups 167 35458.095 212.324 4.809 < 0.001 Within Groups 672 29670 44.152 Non-cumulative Germination – Scarified) Temperature * No. of Days Between Groups 167 63189.524 378.38 52.644 < 0.001 Within Groups 672 4830 7.188 Non-cumulative Germination – Non-scarified) Temperature * No. of Days Between Groups 167 448.095 2.683 1.106 0.196 Within Groups 672 1630 2.426

Texas Tech University, Pablo Teveni, August 2017

100

35/25 Scarified 30/20 Scarified

60 25/15 Scarified 20/10 Scarified

35/25 Non-Scarified 30/20 Non-Scarified

40 Cumulative Germination (%) Germination Cumulative

25/15 Non-Scarified 20/10 Non-Scarified

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

Days in Germination Chamber

Figure 15. Cumulative germination curves for Scarified and Non-Scarified huisache seeds at four different temperature regimes (high/low, °C). Bars represent ± S.E.

Texas Tech University, Pablo Teveni, August 2017

Table 11. Cumulative germination per day*temperature treatment for Scarified huisache seeds. Statistical significance determined using ANOVA with completely randomized design and Duncan’s Multiple Range test. 35/25 Scarified 30/20 Scarified 25/15 Scarified 20/10 Scarified Day 푥̅ Std. Sig. 푥̅ Std. Sig. 푥̅ Std. Sig. 푥̅ Std. Sig. Err Err Err. Err 1 0 0 v◊ 0 0 v 0 0 v 0 0 v 2 0 0 v 0 0 v 0 0 v 0 0 v 3 25 4.7434 r 4 2.9155 tuv 14 2.9155 s 0 0 v 4 81 3.6742 klm 62 1.2247 op 84 3.6742 hijklm 0 0 v 5 91 3.6742 abcdefghi 68 1.2247 no 95 1.5811 abcde 0 0 v 6 92 4.062 abcdefgh 72 2.5495 n 96 1.8708 abcd 0 0 v 7 92 4.062 abcdefgh 79 3.6742 m 97 1.2247 abc 2 1.2247 uv 8 93 3.2912 abcdefg 79 3.6742 m 98 1.2247 ab 5 1.5811 tuv 9 93 3.3912 abcdefg 83 2.5495 ijklm 98 1.2247 ab 8 3.3912 stu 10 93 3.3912 abcdefg 83 2.5495 ijklm 98 1.2247 ab 11 3.6742 st 11 93 3.3912 abcdefg 84 2.9155 hijklm 98 1.2247 ab 13 3.3912 s 12 93 3.3912 abcdefg 85 2.2361 ghijklm 99 1 a 23 2.5495 r 13 93 3.3912 abcdefg 85 2.2361 ghijklm 99 1 a 40 1.5811 q 14 93 3.3912 abcdefg 85 2.2361 ghijklm 99 1 a 56 2.4495 p 15 94 2.9155 abcdef 85 2.2361 ghijklm 99 1 a 61 2.9155 op 16 94 2.9155 abcdef 85 2.2361 ghijklm 99 1 a 68 3.3912 no 17 95 2.2361 abcde 87 3 efghijkl 99 1 a 80 3.1623 lm 18 95 2.2361 abcde 89 2.4495 cdefghij 99 1 a 82 2.5495 jklm 19 95 2.2361 abcde 89 2.4495 cdefghij 99 1 a 86 2.9155 fghijklm 20 95 2.2361 abcde 89 2.4495 cdefghij 99 1 a 88 3 defghijk 21 95 2.2361 abcde 90 2.7386 bcdefghij 99 1 a 89 3.6742 cdefghij 22 95 2.2361 abcde 90 2.7486 bcdefghij 99 1 a 91 3.3166 abcdefghi 23 95 2.2361 abcde 90 2.7386 bcdefghij 99 1 a 91 3.3166 abcdefghi 24 95 2.2361 abcde 90 2.7386 bcdefghij 99 1 a 92 3.3912 abcdefgh 25 95 2.2361 abcde 91 1.8708 abcdefghi 99 1 a 93 2.5495 abcdefg 26 95 2.2361 abcde 92 2 abcdefgh 99 1 a 93 2.5495 abcdefg 27 95 2.2361 abcde 92 2 abcdefgh 99 1 a 93 2.5495 abcdefg 28 95 2.2361 abcde 93 2.5495 abcdefg 99 1 a 93 2.5495 abcdefg 29 95 2.2361 abcde 94 2.9155 abcdef 99 1 a 94 1.8708 abcdef 30 95 2.2361 abcde 94 2.9155 abcdef 99 1 a 94 1.8708 abcdef 31 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 94 1.8708 abcdef 32 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 95 1.5811 abcde 33 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 95 1.5811 abcde 34 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 95 1.5811 abcde 35 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 96 1.8708 abcd 36 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 96 1.8708 abcd 37 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 96 1.8708 abcd 38 96 2.4495 abcd 94 2.9155 abcdef 99 1 a 96 1.8708 abcd 39 96 2.4495 abcd 95 3.1623 abcde 99 1 a 97 2 abc 40 96 2.4495 abcd 96 2.9155 abcd 99 1 a 97 2 abc 41 96 2.4495 abcd 96 2.9155 abcd 99 1 a 97 2 abc 42 96 2.4495 abcd 96 2.9155 abcd 99 1 a 97 2 abc ◊ Mean cumulative germination percentages followed by same letter not significantly different (P=0.05)

Texas Tech University, Pablo Teveni, August 2017

Linear regression analysis revealed that seedling height significantly predicted seedling resprouting percentage (B = 0.001, t(248) = 6.405, P < 0.001). Seedling height also explained a significant proportion of variance in seedling resprouting percentage (R2

= 0.142, F(1,248) = 41.02. P < 0.001). Seedling dry weight also significantly predicted seedling resprouting percentage (B = -0.138, t(128) = -6.84, P < 0.001), and seedling dry weight explained a significant proportion of variance in seedling resprouting percentage

(R2 = 0.268, F(1,128) = 46.79, P < 0.001).

Seedling cotyledon latent meristem heights were significantly different among seedlings of different ages (F(11,108) = 6.545, P < 0.001; Table 13). Average seedling cotyledon latent meristem height of bimonthly samples of ten seedlings was not at or below soil level until 24 months of age (M = -0.8mm, SD = 2.3476), and the greatest cotyledon latent meristem heights were found at 4 months of age (M = 10.1. SD =

4.8178; Figure 16).

According to curvilinear estimation, the best fit for data of seedling cotyledon latent meristem height vs. seedling age (month) was a quadratic function (R2 = 0.287,

F(2,127) = 25.537, P < 0.001, Figure 16), which is as follows:

y = -0.0395x2+ 0.6955x + 4.1846

According to this quadratic function, cotyledon latent meristem height increased to 7.2 mm at month 9, then decreased to 3.9 mm by month 18 and was buried below soil level by month 22.3.

Linear regression analysis revealed that seedling height significantly predicted seedling cotyledon latent meristem height (B = -0.02, t(118) = -6.438, P < 0.001).

Texas Tech University, Pablo Teveni, August 2017

Table 12. Fiducial confidence interval table for dependent variable Seedling Resprouting (%), with covariate plant age (in months), determined using Probit analysis. Probability 95% Confidence Limits Estimate Lower Bound Upper Bound 0.01 5.847 3.521 7.631 0.02 6.765 4.36 8.542 0.03 7.421 4.991 9.18 0.04 7.956 5.523 9.695 0.05 8.42 5.995 10.138 0.06 8.835 6.428 10.534 0.07 9.217 6.831 10.896 0.08 9.572 7.211 11.233 0.09 9.907 7.575 11.552 0.10 10.226 7.923 11.855 0.15 11.659 9.52 13.234 0.20 12.94 10.961 14.515 0.25 14.15 12.301 15.801 0.30 15.333 13.559 17.159 0.35 16.518 14.745 18.639 0.40 17.726 15.875 20.277 0.45 18.98 16.971 22.102 0.50 20.299 18.059 24.145 0.55 21.711 19.164 26.449 0.60 23.246 20.315 29.074 0.65 24.947 21.544 32.122 0.70 26.874 22.891 35.701 0.75 29.121 24.414 40.067 0.80 31.845 26.206 45.601 0.85 35.343 28.436 53.07 0.90 40.296 31.486 64.287 0.91 41.593 32.267 67.342 0.92 43.039 33.135 70.829 0.93 44.709 34.116 74.875 0.94 46.639 35.244 79.672 0.95 48.942 36.574 85.523 0.96 51.793 38.197 92.956 0.97 55.526 40.289 102.994 0.98 60.91 43.242 118.051 0.99 70.474 48.331 146.411

Texas Tech University, Pablo Teveni, August 2017

Table 13. Single-factor analysis of variance table for dependent variable Cotyledon Latent Meristem Height (mm), with fixed factor No. of Months (in age). Source of Degrees Type III Mean F-Statistic Significance Variation of Sum of Square Freedom Squares Between Groups 11 1251.167 113.742 6.545 < 0.001 Within Groups 108 1876.8 17.378

Seedling height also explained a significant proportion of variance in seedling cotyledon latent meristem height (R2 = 0.26, F(1,118) = 41.444. P < 0.001). Seedling dry mass also significantly predicted cotyledon latent meristem height (B = -0.162, t(118) = -8.421, P <

0.001) and explained a significant proportion of variance in seedling cotyledon latent meristem height (R2 = 0.375, F(1,118) = 70.919, P < 0.001).

12 Cotyledon Height (mm) 10 de de Quadratic (Cotyledon Height) cde cde cde 8 e bcd 6 abc abc ab 4 a◊ a 2

Cotyledon Height (mm) Height Cotyledon a 0

-2 0 2 4 6 8 10 12 14 16 18 20 22 24 Seedling Age (in Months) Figure 16. Huisache seedling cotyledon latent meristem height above soil level, measured every other month for 24 months and curvilinear estimation for cotyledon latent meristem height above soil level vs. month. Statistical significance determined using ANOVA with completely randomized design and Duncan’s Multiple Range test. ◊ Means with same letter are not significantly different at P = 0.05 Quadratic (Cotyledon Height): y = -0.0395x2+ 0.6955x + 4.1846; (R2 = 0.287, F(2,127) = 25.537, P < 0.001)

Texas Tech University, Pablo Teveni, August 2017

Probit analysis average marginal effects indicated that month significantly predicted percent of seedlings with cotyledon latent meristems at or below soil level (M =

5.208, SE = 1.199, 95% CI [2.858, 7.558], n = 120, z-score = 4.344, P<0.001).

According to the probit model the percentage of seedlings with cotyledon latent meristems at or below soil level was 1% at 7.3 months (95% CI [1.793, 10.458]), 10% at

11.7 months (95% CI [5.873, 14.391]), and 50% at 20.5 months (95% CI [17.102,

31.332]; Table 14).

Linear regression analysis revealed that seedling height significantly predicted percent of seedlings with cotyledon latent meristems at or below soil level (B = 0.002, t(118) =7.184, P < 0.001). Seedling height also explained a significant proportion of variance in percent of seedlings with cotyledon latent meristems at or below soil level (R2

= 0.304, F(1,118) = 51.604. P < 0.001). Seedling dry mass also significantly predicted percent of seedlings with cotyledon latent meristems at or below soil level (B = 0.013, t(118) = 8.127, P < 0.001) and explained a significant proportion of variance in that parameter (R2 = 0.359, F (1,110) = 66.042, P < 0.001).

Texas Tech University, Pablo Teveni, August 2017

Table 14. Fiducial confidence interval table for dependent variable Cotyledon Latent Meristem Below Soil (%), with covariate plant age (in months), determined using Probit analysis. Probability 95% Confidence Limits Estimate Lower Bound Upper Bound 0.01 7.346 1.793 10.458 0.02 8.286 2.453 11.322 0.03 8.945 2.99 11.918 0.04 9.474 3.468 12.395 0.05 9.928 3.911 12.803 0.06 10.332 4.329 13.168 0.07 10.699 4.731 13.502 0.08 11.039 5.121 13.813 0.09 11.357 5.501 14.108 0.10 11.658 5.873 14.391 0.15 12.993 7.66 15.709 0.20 14.161 9.366 17.012 0.25 15.247 10.992 18.443 0.30 16.293 12.504 20.127 0.35 17.327 13.865 22.18 0.40 18.368 15.066 24.687 0.45 19.435 16.132 27.714 0.50 20.545 17.102 31.332 0.55 21.718 18.016 35.646 0.60 22.98 18.909 40.824 0.65 24.361 19.813 47.123 0.70 25.905 20.759 54.958 0.75 27.683 21.796 65.015 0.80 29.806 22.946 78.533 0.85 32.487 24.338 98.041 0.90 36.205 26.165 129.835 0.91 37.165 26.62 138.979 0.92 38.237 27.122 149.655 0.93 39.542 27.682 162.356 0.94 40.854 28.319 177.837 0.95 42.514 29.061 197.324 0.96 44.55 29.952 222.993 0.97 47.188 31.081 259.212 0.98 50.938 32.64 316.701 0.99 57.462 35.242 434.465

Texas Tech University, Pablo Teveni, August 2017

Chapter 4

DISCUSSION

Study 1: Effective Herbicide Control of Huisache

Month was the most important parameter influencing huisache herbicide mortality across study sites, with two optimal windows of control occurring in May and

October/November. Meyer et al. (1976) found similar results using soil-applied picloram granules, with herbicide-induced mortality highest from March to June and October to

January, lowest during the month of July, and intermediate during the months of

September, August, and February. The results also support the herbicide recommendation offered by Hanselka and Lyons (2010), who instruct treating with picloram + 2,4-D from September through November. Conversely, Bovey et al. (1972) found that greatest canopy reduction occurred when huisache were sprayed with picloram between mid-June and late August; in that study, however, the plants were nursery- grown, and in the previous year of that study the greatest canopy reduction occurred when huisache were sprayed with picloram during mid-September.

Although root crown total nonstructural carbohydrate (TNC) has been positively correlated with herbicide susceptibility in mesquite and other woody rangeland species

(Wan and Sosebee 1990; Sosebee and Dahl 1991; Sosebee 2004), a positive correlation

(R2 = 0.2165) between Future Delta-TNC (average following month TNC – average current month TNC) and huisache mortality was only part of the best-fit model for plants in the partially-defoliated phenological stage. However, a negative correlation between

Future Delta-TNC and huisache mortality was one component of the best-fit models for one of the months of high herbicide-induced mortality. Also, a positive correlation

Texas Tech University, Pablo Teveni, August 2017 between Delta-TNC (average current month TNC – average previous month TNC) and huisache mortality was only part of the best-fit model for the data from the predominantly clay sites, although this was not a particularly strong linear relationship (R2 = 0.045).

The best-fit model for the data from the predominantly sandy loam / sandy clay loam sites included a weakly negative linear relationship (R2 = 0.003) between herbicide mortality and Delta-TNC. These results suggest that using the monthly change in huisache root crown TNC as an indicator for mortality success rate depends upon plant phenology, time of year, and soil texture.

Another possible reason why there were conflicting results when correlating TNC concentration with herbicide-induced mortality is that root crown TNC concentration may not have been the most appropriate metric for quantifying seasonal carbohydrate flux. For example, according to Cralle and Bovey (1996), increases in total nonstructural carbohydrate (TNC) mass in mesquite roots was more important for carbohydrate storage than changes in TNC concentration of roots. Further investigation should be made into how differences in soil texture, phenology, and time of year affect both root crown TNC concentration and herbicide-induced mortality in huisache.

The results from the present chemical control demonstration study supported the results of other studies finding that herbicide treatment can vary according to soil water content, soil temperature, soil type, and plant phenological stage (Fisher et al. 1956; Dahl et al. 1971; Scifres 1980). Bovey et al. (1972) found increased herbicide-induced huisache canopy reduction during the late afternoon, compared with morning and midday applications, which may have been influenced by higher soil temperatures, diurnal changes in leaf physiology, soil water content, or plant water status.

Texas Tech University, Pablo Teveni, August 2017

Soil texture and soil temperature have been found to have an influence on herbicide control of plants such as mesquite (Prosopis glandulosa), as in studies showing more effective control of plants growing in sandy loam, compared to clay soils (Welch undated, Fisher et al. 1956, Fisher et al. 1959). The current study’s findings of more effective control at sites with predominantly clay soils may be related to a differential effect of soil texture upon huisache herbicide susceptibility (compared to the herbicide susceptibility of mesquite) or it may be a result of a relatively small number of experimental sites representing each soil type. Conversely, it may also be related to the complexity of the soils at the different study sites. For example, although the dominant soil types at the Victoria Co. study site (Diebel Ranch) were loamy [Nada/Cieno complex

(sandy loam / sandy clay loam / clay loam) and Telferner fine sandy loam (sandy clay and sandy clay loam subsoils)], a clay pan was present below the soil surface, soils were poorly drained with slow infiltration, and large portions of the rangeland study site were periodically inundated with several centimeters of water (USDA 2017; personal observation). Likewise, although the soils at the Kleberg Co. study site (King Ranch) site were also loamy (sandy loam / sandy clay loam / clay loam / loam) and the soils were well-drained or moderately well-drained, the permeability of the soils was moderately slow or moderate, causing temporary flooding following heavy rains at the study site

(USDA 2017; personal observation). The dominant soil types at the Refugio and San

Patricio study sites also had slow permeability and caused periodic, temporary light flooding (USDA 2017; personal observation). Therefore, all sites in this study were either poorly drained or had slow infiltration, causing periodic shallow innundation, and perhaps obfuscating the effects of soil type upon herbicide-induced mortality.

Texas Tech University, Pablo Teveni, August 2017

As found in mesquite (Sosebee 2004; Dahl et al. 1971), soil temperature at 0.3 m was one of the parameters with the largest effect upon herbicide-induced huisache mortality. However, unlike mesquite, in which herbicide mortality rapidly increases as soil temperature (at 0.3 m) increases from 20 to 28° C (Sosebee 2004; Dahl et al. 1971), the relationship between huisache herbicide mortality and soil temperature is best expressed by a quadratic function with maximum mortality occurring around 24.5°C at

0.3 m. This ideal temperature was similar to the ideal temperature range (between 21.1 and 23.9° C at 30 to 46 cm) described for the herbicide treatment of Texas brush species

(Hanselka 1991, Welch undated). The peak temperature from the current study is also similar to the recommendation given by Hanselka and Lyons (2010), who instruct to spray huisache in autumn, prior to soil temperatures dropping below 23.9° C. Soil temperature appears to have been one of the primary factors influencing the parameter

“month”, which was the most important parameter in AIC analysis, as evident by average soil temperatures between 23.5 and 25° C during the high-mortality months of May and

October, respectively (November, also a high-mortality month, had a lower average soil temperature of about 18.7° C). In the present study the ideal temperature at which to maximize herbicide mortality in huisache was somewhat higher for those plants growing at the predominantly sandy loam / sandy clay loam sites, compared with those growing at the predominantly clay sites (25.22 vs 23.85°C, respectively), possibly due to influence of soil texture upon soil moisture retention, soil pore space, and indirectly, soil temperature.

Phenological stage was another important parameter with a large effect upon herbicide-induced huisache mortality. The results support findings that huisache and other brush species are most susceptible to herbicide mortality when treated during the 71

Texas Tech University, Pablo Teveni, August 2017 full canopy phenological stage, due to the importance of fully-expanded, healthy leaves for herbicide absorption and translocation (Bovey and Meyer 1989; Hanselka and Lyons

2010; Allred 1949; Dahl and Sosebee 1984; Lyons 1998, Welch undated, Fick and

Sosebee 1981). Full canopies of fully-expanded, healthy leaves probably have the highest maximum daily photosynthetic rate, which in mesquite has been found to be highly correlated (R = 0.89 to 0.92) with herbicide control (Meyer et al. 1983).

Soil water content also appeared to be an important parameter, although a greater fit was found more frequently by using its indirect measurement through cumulative precipitation, probably due to missing weather data caused by commonly-occurring malfunctions with the weather stations’ soil water content sensors. Supported by the findings of Bovey et al. (1972), results generally show a positive correlation between herbicide-induced mortality and soil available water content (or cumulative precipitation), which was a component of the best-fit models for herbicide mortality at sandy loam / sandy clay loam sites and during months of high mortality, analyzed separately (Figures A-2 to A-5). The best-fit models during the months of high mortality also included a weak negative correlation between mortality and cumulative precipitation, which may have been a spurious result (caused by small sample size), or it may be that during the months of high herbicide mortality, plants respond differentially to environmental stimulus in regard to herbicide susceptibility.

Further investigation should be made into how soil texture, soil temperature, and soil water content influence herbicide mortality in huisache, especially during the high- mortality months of May, October, and November.

Texas Tech University, Pablo Teveni, August 2017

Study 2: Huisache Seed Germination

Germination percentage of huisache seeds in other studies approximated this study (Jurado et al. 2000; Vora 1989). Germination rates in the scarified 25/15°C treatment were similar to Zisner (1999), who found that scarified huisache seeds grown with bottom heat (22 to 24 degrees C) germinated at 5 days on average. Although other researchers (Vora 1989; Scifres 1974) found the highest germination percentages of huisache seed to occur following 45, 60, 90, or 120 minutes of sulfuric acid scarification, huisache seeds in this study had high germination percentages following concentrated sulfuric acid scarification for 25 minutes, similar to Bush (2008). However, because the seeds were sourced from an unknown provenance in south Texas, the seed characteristics may have differed somewhat from seeds of huisache plants growing in the Coastal Bend region of south Texas. Although final germination percentages of all scarified temperature treatments were statistically the same, germination of seeds in the 25/15°C scarified treatment and 30/20°C scarified treatment lagged behind those of the other two treatments. Because the 35/25-Nonscarified and the 25/15-Nonscarified treatments had germination rates slightly (though not significantly) faster than that of 30/20-

Nonscarified; an unknown factor may have negatively influenced germination rate in the

30/20 temperature treatments. Results showed that when high/low temperatures drop from 25/15°C to 20/10°C, there was a significant decrease in the speed of scarified huisache seed germination. Jurado et al. (2000) also found low temperatures to decrease the cumulative germination rate of scarified huisache seeds at 60 days, with 97% germination at 28°C compared with 68% germination at 12°C. Scifres (1974) found both low temperatures and high temperatures to decrease the cumulative germination rate of 73

Texas Tech University, Pablo Teveni, August 2017

huisache seeds at 10 days, with 75% germination at 30°C, compared with 45% germination at 21°C, 5% germination at 16°C, and 15% germination at 38°C). Although a slower rate of seed germination may not translate into fewer seeds germinating under field conditions, a slower rate of germination allows more time for huisache seeds and seedlings to be consumed or otherwise destroyed by granivores, herbivores, and pathogens.

Study 3: Huisache Seedling Resprouting and Cotyledon Latent Meristem Positioning

Seedlings had to be much older to survive top removal in the current study than in the study by Bovey and Meyer (1974), who found that 76% of huisache seedlings survived top removal at eight weeks old, compared with 22% survival at four weeks old.

That study, however, removed huisache tops within 5 mm of soil level (rather than exactly at soil level) and no description of seed planting depth was provided (which may have affected cotyledon latent meristem positioning following emergence).

The age of huisache seedlings when average cotyledon latent meristem height was at or below soil level (22.3 months) and when over half of plants have cotyledon latent meristems at or below soil level (20.5 months) was similar to the age of huisache seedlings when over half of plants resprouted following top removal (20.3 months) in the resprouting study. No previous studies had been performed on the rate of movement of huisache cotyledon latent meristems over time, though Phil Morey had previously done similar research with mesquite (Pers. comm. R.E. Sosebee 2017).

In the current experiment average huisache seedlings heights were comparable to

(though somewhat less than) huisache seedling heights in other studies (Bush 2008; Vora

Texas Tech University, Pablo Teveni, August 2017

1989; data not shown). However, Bush (2008) did not randomly select seedlings, but instead selected them for uniform size and vigor, and Vora (1989) irrigated plants every other day (instead of every day) and used slow-release fertilizer (instead of using fertilizer dissolved in water). In the present experiment, average huisache seedling masses were significantly lower than those of Polley et al. (1997). However, in that study the soil media contained nitrogen-fixing Rhizobium bacteria and the plant containers were much larger.

Conclusions

Based upon the results from above studies, a number of conclusions can be drawn concerning the ecophysiological and phenological properties of huisache, and their importance for the management of this invasive species. The influence these factors have upon the herbicide susceptibility of huisache had not been examined by previous research. The greatest herbicide control for huisache occurred during the months of

October, May, and November. These results are very similar to the huisache treatment recommendations offered by DOW AgroSciences for South Texas (Pers. comm. B.

Martinez 2012). Disregarding month (which can have highly variable meterological conditions from year to year), the greatest herbicide control occurred as average soil temperatures approached an optimum of 24.5°C and plants were in the full canopy phenological stage. When analyzed separately for month (May, October, or November), soil type (clay or sandy loam / sandy clay loam), and phenology (full canopy or partial defoliation), herbicide control levels responded differentially, with the best models including soil available water content (AWC), cumulative 4-week or 8-week

Texas Tech University, Pablo Teveni, August 2017 precipitation, Delta-TNC, Future Delta-TNC, month, phenology, and/or soil temperature.

Predominantly sandy loam / sandy clay loam sites had highest mortality in the months of

November, October, and May, when plants were in the full canopy phenological stage and average root crown TNC was lower than in the previous month (when month was removed as a parameter, the highest mortality was achieved when soil temperature at 0.3 m was 25.22°C and cumulative rainfall in the previous 8 weeks was high).

Predominantly clay sites, meanwhile, had the highest huisache mortality when soil temperatures were 23.85°C and average root crown TNC was higher than in the previous month. These results suggest that temporary changes in huisache herbicide susceptibility occur in response to differences in phenology, root crown physiology, soil environment, and time of year. These results also provide direction for future research into the herbicide susceptibility of huisache. The influence of soil type, phenological stage

(especially full canopy and partial defoliation), and month (especially October, May, and

November) should be more carefully and more precisely studied in order to better assist land managers in describing the optimum conditions for herbicide management of huisache under more geographically and environmentally varied conditions. In addition, because the current study used individual plant treatment application of herbicide, further research will be needed to relate these findings to aerial application, which is a much more common method for applying herbicide to control woody rangeland pest species.

Based upon research results, in order to achieve the highest rates of herbicide- induced huisache mortality, ranchers and other land managers generally should spray huisache during the full canopy phenological stage when soil temperature at 0.3 m is at or near 24.5°C (usu. during the months of May, October, or November). However, in

Texas Tech University, Pablo Teveni, August 2017 predominantly sandy loam / sandy clay loam soils, huisache should be sprayed when soil temperature is a little higher (at or near 25.22°C) and when recent (previous 8 weeks) precipitation has been high. Conversely, in predominantly clay soils, huisache should be sprayed when soil temperature is at or near 23.85°C.

The current study also found that acid scarification increased huisache seed germination by roughly fivefold. Previous huisache seed germination studies had not used diurnally-fluctuating temperature ranges. Temperatures between 35° and 10°C had no negative impact on total seed germination, although plants in the 20/10°C treatment had significantly slower germination. This means that huisache can proliferate through seed production over a wide range of soil temperatures.

Previous research had not been conducted on the burying of the cotyledon latent meristem in huisache, and little research had been done on huisache seedling resprouting ability. However, the current study found that ≥50% of huisache seedlings re-sprouted at

20.3 months of age, average cotyledon latent meristem height was below soil level at

22.3 months of age, and ≥50% cotyledon latent meristems sank to below soil level at 20.5 months of age. This means that most huisache plants will survive brush treatments consisting of top-removal when they are less than 2 years old

The results from this research provide guidance for land managers to better understand and control the growth and spread of the invasive rangeland brush species huisache (Acacia farnesiana), resulting in greater herbicide efficacy, lower overall cost of management, greater forage availability for livestock, improved habitat quality for livestock and wildlife, as well as other benefits. These research results also point toward specific directions where future research can be conducted in order to gain a more

Texas Tech University, Pablo Teveni, August 2017 complete understanding of the physiology and management of this invasive rangeland species.

Texas Tech University, Pablo Teveni, August 2017

LITERATURE CITED

Actkinson MA, Kuvlesky WP Jr., Boal CW, Brennan LA, and Hernández F. 2007.

Nesting habit relationships of sympatric crested caracaras, red-tailed hawks, and

white-tailed hawks in South Texas. Wilson J Ornith. 119(4): 570-578.

Allison DV and Rechenthin CA. 1956. Root plowing proved best method of brush

control in south Texas. J. Range Manage. 9: 130-133.

Allred BW. 1949. Distribution and control of several woody plants in Texas and

Oklahoma. J. Range Manage 2(1): 17-29.

Amundson RG, Ali AR, and Belsky AJ. 1995. Stomatal responsiveness to changing light

intensity increases rain-use efficiency of below-crown vegetation in tropical

savannas. J. Arid. Environ. 29: 139-53.

Andersson M, Michelsen A, Jensen M, and Kjøller A. 2004. Tropical savannah

woodland: Effects of experimental fire on microorganisms and soil emissions of

carbon dioxide. Soil Biol and Biochem. 36: 849-858.

Archer S. 1989. Have southern Texas savannas been converted to woodlands in recent

history? Am. Nat. 134: 545-561.

Archer S. 1990. Development and stability of grass/woody mosaics in a subtropical

savanna parkland, Texas, U.S.A. J Biogeography. 17(4/5): 453-462.

Archer S. 1994. Woody plant encroachment into southwestern grasslands and savannahs:

rates, patterns, and proximate causes, In: Vavra M, Laycock WA, and Pieper RD,

editors. Ecological implications of livestock herbivory in the west. Denver (CO):

Society for Range Management. p. 13-68.

Archer S. 1995a. Harry Stobbs memorial lecture, 1993: Herbivore mediation of grass- 79

Texas Tech University, Pablo Teveni, August 2017

woody plant interactions. Tropical Grasslands. 29: 218-235.

Archer S. 1995b. Tree-grass dynamics in a Prosopis-thornscrub savanna parkland:

Reconstructing the past and predicting the future. Ecoscience. 2(1): 83-99.

Archer S. 1996. Assessing and interpreting grass-woody plant dynamics. In: Hodgson J

and Illius A, editors. The ecology and management of grazing systems.

Wallingford, Oxon (UK): CAB International. p. 101-134.

Archer S, Boutton TW, and Hibbard KA. 2000. Trees in grasslands: Biochemical

consequences of woody plant expansion. In: Schulze ED, Harrison SP, Heimann

M, Holland EA, Lloyd J, Prentice IC, Schimel D, editors. Global biochemical

cycles in the climate system. San Diego (CA): Academic Press. p. 115-139.

Archer S, Scifres C, Bassham CR, and Maggio R. 1988. Autogenic succession in a

subtropical savannah: conversion of grassland into thorn woodland. Ecol Monogr.

58: 111-127.

Backéus I. 1992. Distribution and vegetation dynamics of humid savannas in Africa and

Asia. J. Veg. Sci. 3: 345-57.

Bade D. 1991. Grazing management following brush control. In: Welch TG, editor.

Proceedings - Brush Management Symposium; 1991 May 10-11; Giddings (TX).

College Station (TX): Texas Agricultural Extension Service. p. 48-54.

Badi KH. 1967. Afforestation on the clay plains of Kassala Province [Sudan]: (A) High

rainfall clay plains. (B) Arid clay plains. Khartoum (Sudan): Forest Department.

20 p. Bulletin No. 13.

Bailey RG. 1980. Description of the ecoregions of the United States. Ogden (UT):

USDA Forest Service. 77 p. Misc. Publ. 1391.

Texas Tech University, Pablo Teveni, August 2017

Bailey V. 1905. Biological survey of Texas. North American Fauna. 25: 1–222.

Barnes PW and Archer S. 1999. Tree-shrub interactions in a subtropical savanna

parkland: Competition or facilitation? J. Veg. Sci. 10:525-536.

Bartlett JR. 1854. Personal narrative of explorations and incidents in Texas, New Mexico,

California, Sonora, and Chihuahua, connected with the United States and

Mexican Boundary Commission during the years 1850, ‘51, ‘52, and ‘53. (2 vol.)

New York: D. Appleton and Co. 1130 p.

Batianoff GN and Butler DW. 2002. Assessment of invasive naturalized plants in

southeast Queensland. Plant Prot Q. 17: 27-34.

Beale IF. 1973. Tree density effects on yields of herbage and tree components in south

west Queensland mulga (Acacia aneura F. Muell.) scrub. Tropical Grasslands.

7:135-142.

Beck DL, RE Sosebee, and EB Herndon. 1975. Chemical control of mesquite regrowth of

different ages. J of Range Manage. 28(5): 408-410

Bedunah DJ, and RE Sosebee. 1984. Forage response of a mesquite-buffalograss

community following range rehabilitation. J. Range Manage. 37(6):483-487.

Belsky AJ. 1994. Influences of trees on savanna productivity: Tests of shade, nutrients,

and tree-grass competition. Ecology 75(4):922-932.

Belsky AJ, Amundson RG, Duxberry RM, Riha SJ, Ali AR, and Mwonga SM. 1989. The

effects of trees on their physical, chemical, and biological environments in a semi-

arid savanna in Kenya. J Appl Ecol. 26:1004-24.

Belsky AJ, Mwonga SM, Amundson RG, Duxbury JM, and Ali AR. 1993a. Comparative

effects of isolated trees on their undercanopy environments in high- and low-

Texas Tech University, Pablo Teveni, August 2017

rainfall savannas. J of Appl Ecol. 30:143- 155.

Belsky AJ, Mwonga SM, and Duxbury JM. 1993b. Effects of widely spaced trees and

livestock grazing on understory environments in tropical savannas. Agroforestry

Systems. 24:1-20.

Berdowski JJM. 1987. Transition from heathland to grassland initiated by the heather

beetle. Vegetatio. 72(3):167-173.

Bernhard-Reversat F. 1982. Biogeochemical cycle of nitrogen in a semi-arid savanna.

Oikos. 38:321-332.

Berlandier JL. 1834. The diary of Jean Louis Berlandier in Texas. (trans.). mimeo. 353 p.

Berlandier JL. 1980. Journey to Mexico during the years 1826 to 1834 (2 vols.). Trans.

Olendorf SM, Bigelow JM, and Standifer MM. The Texas State Historical

Association and Center for Studies in Texas History, University of Texas: Austin.

Bertness MD and Callaway R. 1994. Positive interactions in communities. Trends Ecol.

Evol. 9: 191-193.

Bogusch ER. 1952. Brush invasion in the Rio Grande Plain of Texas. Texas J. Sci. 4: 85-

91.

Bond WJ and GF Midgley. 2000. A proposed CO2-controlled mechanism of woody plant

invasions in grasslands and savannas. Global Change Biol. 6: 865-869.

Bontrager OE, CJ Scifres, and DL Drawe. 1979. Huisache control by power grubbing. J

Range Manage. 32(3):185-188.

Boutton TW, Archer SR, Midwood AJ, Zitzer SF, and Bol R. 1998. δ13C values of soil

organic carbon and their use in documenting vegetation change in a subtropical

savanna ecosystem. Geoderma. 82: 5-41.

Texas Tech University, Pablo Teveni, August 2017

Bovey RW, Bower JR, and Morton HL. 1970. Control of huisache and associated woody

species in south Texas. J Range Manage. 23: 47-50.

Bovey RW, Davis FS, Merkle MG. 1967. Distribution of picloram in huisache after foliar

and soil applications. Weeds. 15(3): 245-249.

Bovey RW, Haas RH, and Meyer RE. 1972. Daily and seasonal response of huisache and

macartney rose to herbicides. Weed Sci. 20(6): 577-580.

Bovey RW and Meyer RE. 1989. Control of huisache and honey mesquite with a

carpeted roller herbicide applicator. J of Range Manage. 42: 407-411.

Bovey RW and Meyer RE. 1974. Mortality of honey mesquite and huisache seedlings

from herbicides and top removal. Weed Sci. 22(3): 276-279.

Bovey RW, Meyer RE, and Whisenant SG. 1990. Effect of simulated rainfall on

herbicide performance in huisache (Acacia farnesiana) and honey mesquite

(Prosopis glandulosa). Weed Technol. 4: 26-30.

Box TW. 1960. Herbage production in four range plant communities in South Texas. J

Range Manage. 13(2): 72-76.

Box TW and J Powell. 1965. Brush management techniques for improved forage values

in South Texas. J Range Manage. 18: 285-296.

Box TW., Powell J, and Drawe DL. 1967. Influence of fire on south Texas chaparral

communities. Ecology 48(6): 955-961.

Branson FA. 1985. Vegetation changes on western rangelands (Range Monograph No.

2). Denver (CO): Society for Range Management. 76 p.

Brown JR, and Archer S. 1989. Woody plant invasion of grasslands: Establishment of

honey mesquite (Prosopis glandulosa var. glandulosa) on sites differing in

Texas Tech University, Pablo Teveni, August 2017

herbaceous biomass and grazing history. Oecologia. 80: 19-26.

Brown JR and Archer S. 1990. Water relations of a perennial grass and seedlings vs adult

woody plants in a subtropical savanna, Texas. Oikos. 57: 366-74.

Burrows WH, Scanlan JC, and Anderson ER. 1988. Plant ecological relations in open

forest, woodlands and shrublands. In: Burrows WH, Scanlan JC, and Rutherford

MT, editors. Native pastures in Queensland: The resources and their management.

Brisbane (Australia): Queensland Dept. of Primary Industries, Queensland

Government Printer. p. 72-90.

Burrows WH, Carter JO, Scanlan JC, and Anderson ER. 1990. Management of savanna

for livestock production in north-east Australia: Contrasts across the tree- grass

continuum. J Biogeogr. 17: 503-512.

Bush JK. 2008. Aboveground and belowground growth of seedlings of an early and late

successional species in infertile and fertile soil. Southwest Nat. 53(1): 39-44.

Bush JK and Van Auken OW. 1990. Growth and survival of Prosopis glandulosa

seedlings associated with shade and herbaceous competition. Bot. Gaz. 151(2):

234-239.

Caldwell MM and Richards JH. 1989. Hydraulic lift: Water efflux from upper roots

improves effectiveness of water uptake by deep roots. Oecologia 79:1-5.

Cantor LF and Whitham TG. 1989. Importance of belowground herbivory: Pocket

gophers may limit aspen to rock outcrop refugia. Ecology 70: 962-970.

Carter CW, and Holt J. Undated. Ecological Site Characteristics: Clay Loam 20-25” PZ

[Internet]. Washington (D.C.): U.S. Department of Agriculture, Natural Resources

Conservation Service. [cited 2016 June 14] Available from:

Texas Tech University, Pablo Teveni, August 2017

https://esis.sc.egov.usda.gov/ESDReport/fsReportPrt.aspx?id=R083CY446TX&r

ptLevel =all&approved=yes&repType=regular&scrns=&comm=

Carter CW, Reinke TW, and Garcia V. 2009. Ecological Site Description: Loamy Prairie

28-40” PZ [Internet]. [cited 2016 May 24]. Washington (D.C): U.S. Dept. of

Agriculture, Natural Resources Conservation Service. Available from:

https://esis.sc.egov.usda.gov/ESDReport/fsReport.aspx?approved=yes&rptLevel=

all&id=R150AY535TX

Céliz FF. 1935. Diary of the Alarcon expedition into Texas, 1718-1719. Trans. F.L.

Hoffman. Los Angeles (CA): Quivira Society Publications. Vol. 5: 124 p.

Ciais, P, Tans PP, Trolier M, White JWC, and Francey RJ. 1995. A large northern

13 12 hemisphere CO2 sink indicated by the C/ C ratio of atmospheric CO2. Science

269: 1098-1102.

Clarkson NM, and Lee GR. 1988. Effects of grazing and severe drought on a native

pasture in the Traprock region of southern Queensland. Tropical Grasslands 2(4):

176-183.

Cohn EJ, Van Auken OW, and Bush JK. 1989. Competitive interactions between

Cynodon dactylon seedlings and Acacia smallii seedlings at different nutrient

levels. Am Midl Nat. 121: 265-272.

Cole KL. 1985. Late Holocene vegetation changes in Greenwater Valley, Mojave Desert,

California. Quaternary Res. 23(2): 227-235.

Cralle HT and Bovey RW. 1996. Total nonstructural carbohydrates and regrowth in

honey mesquite (Prosopis glandulosa) following hand defoliation or clopyralid

treatment. Weed Sci. 44: 566-569.

Texas Tech University, Pablo Teveni, August 2017

Crisp MD and Lange RT. 1976. Age structure, distribution, and survival under grazing of

the arid-zone shrub Acacia burkittii. Oikos 27:86-92.

Crosswhite FS. 1980. Dry country plants of the south Texas plains. Desert Plants 2: 141-

179.

Dacy EC and Fulbright TE. 2009. Survival of sprouting shrubs following summer fire:

effects of morphological and spatial characteristics. Range Ecol and Manage. 62:

179-185.

Dahl BE and Sosebee RE. 1984. Timing - the key to herbicidal control of mesquite.

Lubbock (TX): Department of Range and Wildlife Management, Texas Tech

University. College of Agricultural Sciences Contribution No. T-9-352.

Dahl BE, Wadley RB, George MR, and Talbot JL. 1971. Influence of site on mesquite

mortality from 2,4,5-T. J of Range Manage. 24: 210-15.

Davis LI. 1934. Distribution of black-throated green warblers and Wilson’s warblers

wintering in Cameron County, Texas, during the season of 1933-1934. Wilson

Bull. 46(4): 223-227.

Davis RB, and Winkler CK. 1968. Brush vs. cleared range as deer habitat in southern

Texas. J Wildlife Manage. 32(2): 321-329.

Dawson TE. 1993. Hydraulic lift and water use by plants -- implications for water

balance, performance, and plant-plant interactions. Oecologia 95: 565-74.

Drawe DL. 1980. The role of fire in the Coastal Prairie. In: Hanselka CW, editor.

Prescribed range burning in the Coastal Prairie and Eastern Rio Grande Plains in

Texas: Proceedings of a Symposium; 1980 Oct 16; Kingsville, TX. College

Station (TX): Texas Agricultural Extension Service. p. 101-113. Contribution No.

Texas Tech University, Pablo Teveni, August 2017

TA 16277.

Drawe DL and Box TW. 1968. Forage ratings for deer and cattle on the Welder Wildlife

Refuge. J of Range Manage. 21(4): 225-228.

Duaine CL. 1971. Caverns of Oblivion. Corpus Christi (TX): Carl L. Duaine. 240 p.

Ebinger JE, Siegler DS, and Clarke HD. 2000. Taxonomic revision of South American

species of the genus Acacia subgenus Acacia (Fabaceae: Mimosoideae). Syst Bot.

25(4): 588-617.

Ebinger JE, Siegler DS, and Clarke HD. 2002. Notes on the segregates of Acacia

farnesiana (L.) Willd. (Fabaceae: Mimosoideae) and related species in North

America. Southwest Nat. 47(1): 86-91.

Eddy MR and Judd FW. 2003. Phenology of Acacia berlandieri, A. minuata, A. rigidula,

A. schaffneri, and Chloroleucon ebano in the Lower Rio Grande Valley of Texas

during a drought. Southwest Nat. 48(3): 321-332.

El-Gamassy AM and Rofaeel IS. 1975. The effect of tree age and time of day for

collecting the flowers on the flower yield, content, and composition of cassie

(Acacia farnesiana) essential oils. Egyptian J of Hort. 2(1):39-52.

El-Lakany MH. 1987. Use of Australian acacias in North Africa. In: Turnbull JW, editor.

Australian acacias in developing countries: Proceedings of an international

workshop (ACIAR Proceedings No. 16); 1986 Aug 4-7; Gympie, Queensland

(Australia). Canberra, Australia: Australian Centre for International Agricultural

Research. p. 116-117.

Engle DM and Bonham CD. 1980. Nonstructural carbohydrates in roots of gambel oak

sprouts following herbicide treatment. J Range Manage. 33(5): 390-394.

Texas Tech University, Pablo Teveni, August 2017

Everitt DH, and Drawe DL. 1993. Trees, shrubs, and cacti of South Texas. 2nd ed.

Lubbock (TX): Texas Tech University Press. 213 p.

Everitt JH, CL Gonzales, Scott G, and Dahl BE. 1981. Seasonal food preferences of

cattle on native range in the South Texas plains. J Range Manage. 34(5): 384-388.

Fick WH, and RE Sosebee RE. 1981. Translocation and storage of 14C-labeled total

nonstructural carbohydrates in honey mesquite. J Range Manage. 34(3): 205-208.

Fisher CE, Meadors CH, and Behrens R. 1956. Some factors that influence the

effectiveness of 2,4,5-trichlorophenoxyacetic acid in killing mesquite. Weed Sci.

4: 139-147

Fisher CE, Meadors CH, Behrens R, Robison ED, Marion PT, and Horton HL. 1959.

Control of mesquite on grazing lands. College Station (TX): Texas Agricultural

Experiment Station. 24 p. Bulletin 935.

Fisher CE, Wiedemann HT, Meadors CH, and Brock JH. 1973. Mechanical control of

mesquite. In: C.J. Scifres, editor. Mesquite: Growth and development,

management, economics, control, uses. College Station (TX): Texas Agricultural

Experiment Station. p. 46-52. Resource Monogram 1.

Forrestal PP. 1934. Pena’s diary of the Aguayo expedition. Preliminary Studies of the

Texas Catholic Historical Society. Vol. 2(7): 68 p.

Frías-Hernández JT, Aguilar-Ledezma AL, Olalde-Portugal V, Balderas-Lopez JA,

Gutiérrez-Juárez G, Alvarado-Gil JJ, Castro JJ, Vargas H, Arbores A, Dendooven

L, Miranda LCM. 1999. Research Note: Soil characteristics in semiarid highlands

in central Mexico, as affected by mesquite trees (Prosopis laevigata). Arid Soil

Res and Rehab. 13: 305-312.

Texas Tech University, Pablo Teveni, August 2017

Fulbright TE. 1991. Why does brush increase? In: Welch TG editor. Proceedings - Brush

Management Symposium; 1991 May 10-11; Giddings (TX). College Station

(TX): Texas Agricultural Extension Service. p. 6-15.

Fulbright TE, Kuti JO, and Tipton AR. 1997. Effects of nurse-plant canopy light intensity

on shrub seedling growth. J Range Manage. 50(6): 607-610.

Fulbright TE, and Bryant FC. 2003. The Wild Horse Desert: climate and ecology. In:

Forgason CA, Bryant FC, and Genho PC, editors. Ranch Management:

Integrating Cattle, Wildlife, and Range. Kingsville (TX): King Ranch Institute. p.

33-58.

Fulbright TE and Taylor RB. 2001. Brush management for white-tailed deer. Kingsville

(TX) and Austin (TX): A Joint Wildlife Technical Publication of the Caesar

Kleberg Wildlife Research Institute, Texas A&M University-Kingsville and

Texas Parks and Wildlife. 24 p.

Geesing D, Felker P, and Bingham RL. 2000. Influence of mesquite (Prosopis

glandulosa) on soil nitrogen and carbon development: Implications for global

carbon sequestration. J Arid Environ. 46: 157-180.

Gill LS, Jagede RO, and Husaini SWH. 1986. Studies on the seed germination of Acacia

farnesiana (L.) Willd. J. Tree Sci. 5(2): 92-97.

Glenn EP, Pitelka LF, and Olsen MW. 1992. The use of halophytes to sequester carbon.

Water Air Soil Poll. 64: 251-263.

Griffith GE, Bryce SA, Omernik JM, Comstock JA, Rogers AC, Harrison B, Hatch SL,

and Bezanson D. 2004. Ecoregions of Texas (color poster with map, descriptive

text, and photographs): Reston (VA): U.S. Geological Survey (map scale

Texas Tech University, Pablo Teveni, August 2017

1:2,500,000).

Guckian WJ. 1988. Soil Survey of Refugio County, Texas. Washington (D.C.): U.S.

Department of Agriculture, Soil Conservation Service. 124 p.

Guckian WJ and Garcia RN. 1979. Soil Survey of San Patricio and Aransas Counties,

Texas. Washington (D.C.): U.S. Department of Agriculture, Soil Conservation

Service. 123 p.

Hamilton WT. Undated. Which brush management practice will work for you? College

Station (TX): Texas Agricultural Experiment Station. p. 226-235.

Hamilton WT. 1991. Putting it together: An effective brush management plan. In: Welch

TG, editor. Proceedings - Brush Management Symposium; 1991 May 10-11;

Giddings (TX). College Station (TX): Texas Agricultural Extension Service. p. 55

to 65.

Hanselka CW. Undated. Maintenance vs. reclamation brush control. Corpus Christi (TX):

Texas Agricultural Extension Service. p. 138-148

Hanselka CW. 1991. Brush management decisions: Ten factors to consider. In: Welch

TG, editor. Proceedings - Brush Management Symposium; 1991 May 10-11;

Giddings (TX). College Station (TX): Texas Agricultural Extension Service. p.

16-22.

Hanselka CW. 2009. Ecological Site Characteristics: Claypan Prairie 28-44 [Internet].

Washington (D.C.): U.S. Department of Agriculture, Soil Conservation Service.

[modified 2009 March 6; cited 2016 July 11]. Available from:

https://esis.sc.egov.usda.gov/ESDReport/fsReportPrt.aspx?id=R150AY528TX&r

ptLevel=all&approved=yes&repType=regular&scrns=&comm=

Texas Tech University, Pablo Teveni, August 2017

Hanselka CW, Drawe DL, and Ruthven DC III. 2004. Management of South Texas

rangelands with prescribed fire. In: Sosebee RE, Wester DB, Britton DB,

McArthur CM, Durant E, and Kitchen SG, editors. Proceedings: Shrubland

dynamics – Fire and water (Proceedings RMRS-P-47); Lubbock (TX); 2004

August 10-12. Fort Collins (CO): USDA, Forest Service, Rocky Mountain

Research Station. p. 57-61.

Hanselka CW, Hamilton WT, and Conner JR. 1996. Integrated brush management

systems (IBMS): Strategies and economics. College Station (TX): Texas A&M

AgriLife Extension, Texas A&M University. E-407. 12 p.

Hanselka CW and Lyons RK. 2010. Brush busters: How to control huisache. College

Station (TX): Texas Agrilife Extension Service. 3 p. L-5348.

Hanselka CW and Reinke T. 2009. Ecological Site Description: Lowland 35-56” PZ

[Internet]. Washington (D.C.): U.S. Department of Agriculture, Natural Resouces

Conservation Service. [cited 2016 May 28]. Available from: https://esis.sc.egov.

usda.gov/ESDReport/fsReportPrt.aspx?id=R150AY537TX&rptLevel=all&approv

ed=yes&repType=regular&scrns=&comm=

Hansmire, JA, Drawe DL, Wester DB, and Britton CM. 1988. Effects of winter burns on

forbs and grasses of the Texas Coastal Prairie. Southwest Nat. 33(3): 333-338.

Harrington GN. 1991. Effects of soil moisture on shrub seedling survival in semi-arid

grassland. Ecology 72: 1138-1149.

Heitschmidt RK, Schulz RD, and Scifres CJ. 1986. Herbaceous biomass dynamics and

net primary production following chemical control of honey mesquite. J Range

Manage. 39: 67-71.

Texas Tech University, Pablo Teveni, August 2017

Herbel CH, Ares EN, and Wright RA. 1972. Drought effects on a semi-desert grassland.

Ecology. 53: 1084-1093.

Herrera-Arreola G, Herrera Y, Reyes-Reyes BG, and Dendooven L. 2007. Mesquite

(Prosopis juliflora (Sw.) DC.), huisache (Acacia farnesiana (L.) Willd.), and

catclaw (Mimosa biuncifera Benth.) and their effects on dynamics of carbon and

nitrogen in soils of the semi-arid highlands of Durango Mexico. J Arid Environ.

69: 583-598.

Hester TR. 1980. Digging Into South Texas Prehistory: A Guide for Amateur

Archaeologists. San Antonio (TX): Corona Publishing Co. 201 p.

Hibbert AR. 1983. Water yield improvement potential by vegetation management on

western rangelands. Water Resour Bull. 19: 375-381.

Hobbs J and Leon FG III. 1987. Great-tailed grackle predation on South Texas citrus

(Identifying a unique problem). In: Holler NR, editor. Proceedings of the Third

Eastern Wildlife Damage Control Conference; 1987 Oct 18-21; Gulf Shores (AL).

Lincoln (NB): University of Nebraska. p. 143-148.

Hoffman GO. 1981. Control and management of mesquite on rangeland. College Station

(TX): Texas Agricultural Extension Service, Texas A&M University. 15 p. B-

1381.

Hoffman WA, Orthen B, and Vargas Do Nascimento PK. 2003. Comparative fire ecology

of tropical savanna and forest trees. Funct Ecol. 17: 720-726.

Idso SB. 1992. Shrubland expansion in the American Southwest. Climate Change. 22:

85-86.

Jensen R. 1988. Changing the face of the range – will brush control boost water yields?

Texas Tech University, Pablo Teveni, August 2017

Texas Water Research Institute. College Station (TX): Texas Water Research

Institute. 10 p. Vol. 14, No. 1.

Johnson L Jr. 1963. Pollen analysis of two archaeological sites at Amistad Reservoir,

Texas. Texas J. Sci. 15: 225-230.

Johnson HB., H.W. Polley, H.S. Mayeux. 1993. Increasing CO2 and plant-plant

interactions: Effects on natural vegetation. Vegetatio 104-105: 157-170.

Johnston MC. 1963. Past and present grasslands of southern Texas and northeastern

Mexico. Ecology. (44): 456-465.

Jones FB. 1975. Flora of the Texas Coastal Bend. Sinton (TX): Rob and Bessie Welder

Wildlife Foundation. Welder Wildlife Contrbution B-6.

Joshi HB. 1983. The Silviculture of Indian Trees. Revised ed.. Delhi (India): Government

of India Press. 344 p. Vol. 4.

Jurado E, Aguirre O, Flores J, Návar J, Villalón H, and Wester D. 2000. Germination in

Tamaulipan Thornscrub of north-eastern Mexico. J Arid Environ. 46: 413-424.

Kanai R, and Black CC. 1972. Biochemical basis for net CO2 assimilation in C4-plants.

In: Black CC, editor. Net carbon dioxide assimilation in higher plants.

Symposium Southern Section American Society Plant Physiologists and Cotton,

Inc.; 1972 Apr 7; Mobile (AL). Raleight (NC): American Society of Plant

Physiologists. p. 75-85.

Kelton E. 1975. The story of Rocky Creek. The Practicing Nutritionist. 9: 1-5.

Kennard DG and BH Walker. 1973. Relationships between tree canopy cover and

Panicum maximum in the vicinity of Fort Victory. Rhod J of Agr Res. 11: 145-

153.

Texas Tech University, Pablo Teveni, August 2017

King JR, Conway WC, Rosen DJ, Oswald BP, and Williams HM. 2014. Total

nonstructural carbohydrate trends in deeproot sedge (Cyperus entrerianus). Weed

Sci. 62: 186-192.

Klemmedson, JO, and Tiedemann AR. 1986. Long-term effects of mesquite removal on

soil characteristics: II. Nutrient availability. Soil Sci Soc of Am J. 50: 476-480.

Knoop, WT, and Walker BH. 1985. Interactions of woody and herbaceous vegetation in

southern African savanna. J. Ecol. 73: 235-53.

Kononoff PJ and Heinrichs AJ. 2003. The effects of corn silage particle size and

cottonseed hulls on cows in early lactation. J. Dairy Sci. 86:2438–2451.

Kunkel G. 1976. Notes on the introduced elements in the Canary Islands’ Flora. In:

Kunkel G, editor. Biogeography and Ecology in the Canary Islands. The Hague

(Netherlands): Dr. W. Junk b.v., Publishers. p. 249-66.

Lawrence BM. 1984. Progress in essential oils. Perfumer and Flavorist. 9(3):35,45.

Lehmann VW. Undated. Quail management: What you can do. In: Wildlife Management

Handbook. College Station: Texas AgriLife Extension, Texas A&M University.

p. III-A 1 to III-A 4.

Lewin R. 1985. Plant communities resist climate change. Science 228: 165-166.

Lonard RI, Everitt JH, and Judd FW. 1991. Woody plants of the Lower Rio Grande

Valley, Texas. Austin (TX): Texas Memorial Museum, University of Texas. 179

p. Miscellaneous Publications, No. 7.

Lyons RK. 1998. Mesquite control using leaf-spray individual plant treatments. Texas

A&M Agricultural Research and Extension Center, Uvalde, TX: 1 p.

Lyons RK, Ginnett TF, and Taylor RB. 1999. Woody plants and wildlife: Brush sculpting

Texas Tech University, Pablo Teveni, August 2017

in South Texas and the Edwards Plateau. College Station: Agrilife Extension,

Texas A&M University System. 6 p. L-5532.

Marion WR and Fleetwood RJ. 1978. Nesting ecology of the plains chachalaca in South

Texas. Wilson Bull. 90(3): 386-395.

Maron, JL and Connors PG. 1996. A native nitrogen-fixing shrub facilitates weed

invasion. Oecologia. 105: 302-312.

Mayeux HS, Johnson HB, and Polley HW. 1991. Global change and vegetation

dynamics. In: James LF, Evans JO, Ralphs MH, and Sigler BJ, editors. Noxious

range weeds. Boulder (CO): Westview Press. p. 62-74.

McCarron JK and Knapp AK. 2001. C3 woody plant expansion in a C4 grassland: Are

grasses and shrubs functionally distinct? Am J Bot. 88(10): 1818-1823.

McGinty A. 1991. Rangeland weed and brush management technologies: Past and

present. In: Welch TG, editor. Proceedings - Brush Management Symposium;

1991 May 10-11; Giddings (TX). College Station (TX): Texas Agricultural

Extension Service. p. 31-38.

McLendon T. 1991. Preliminary description of the vegetation of South Texas exclusive

of coastal saline zones. Texas J Sci. 43: 13-32.

McPherson GR. 1993. Effects of herbivory and herb interference on oak establishment in

a semi-arid temperate savanna. J of Veg Sci. 4: 687-692.

Meyer RE and Bovey RW. 1982. Establishment of honey mesquite and huisache on a

native pasture. J Range Manage. 35(5): 548-550.

Meyer RE, Bovey RW, Riley TE, and McKelvy WT. 1972. Leaf removal interval effects

after sprays to woody plants. Weed Sci. 20: 498-501.

Texas Tech University, Pablo Teveni, August 2017

Meyer RE, RW Bovey, TE Riley, and TO Flynt. 1976. Seasonal response of macartney

rose and huisache to herbicides. J Range Manage. 29(2): 157-160.

Meyer RE, Hanson JD, and Dye AJ. 1983. Correlation of honey mesquite response to

herbicides with three plant variables and soil water. J. Range Manage. 36(5): 613-

615.

Meyer JY. 2000. Invasive species in the Pacific: A technical review and draft regional

strategy. In: Sherley G, editor. Samoa: South Pacific Regional Environment

Programme. p. 93.

Miller WL. 1982. Soil Survey of Victoria County, Texas. Washington (D.C.): U.S.

Department of Agriculture, Soil Conservation Service. 149 p.

Mordelet P and JC Menaut. 1995. Influence of trees on above-ground production

dynamics of grasses in a humid savanna. J Veg Sci. 6: 223-28.

Murphy RP. 1958. Extraction of plant samples and the determination of total soluble

carbohydrates. J Sci Food Agr. 9: 714-717.

Muttiah RS and Johnson HB. 1999. A biomass model without photorespiration for

huisache at different CO2 levels. Biotronics. 28: 23-31.

Mutz JL, Scifres CJ, Drawe DL, Box TW, and Whitson RE. 1978. Range vegetation after

mechanical brush treatments on the Coastal Prairie. Texas Agricultural

Experiment Station Bulletin 1191. 18 p.

Naranjo A, Salas M, Agudo L, Fernandez E, and Arévalo JR. 2009. Studies on the

distribution and characteristics of an allochthonous population of Acacia

farnesiana. Open Forest Sci J. 2: 91-97.

Návar J and Bryan R. 1990. Interception loss and rainfall redistribution by three semiarid

Texas Tech University, Pablo Teveni, August 2017

growing shrubs in northeastern Mexico. J. Hydrol. 115: 51-63.

Newcomb WW Jr. 1961. The Indians of Texas. Austin: University of Texas Press. 404 p.

Neilson RP. 1986. High resolution climatic analysis and southwest biogeography.

Science 232: 27-34.

Núñez Cabeza de Vaca A. 1871. Relation of Alvar Nuñez Cabeza de Vaca. T.B. Smith,

trans. New York: J. Munsell. 300 p.

Parrotta JA, 1992. Acacia farnesiana (L.) Willd. aroma, huisache, Leguminoseae

(Mimosoideae), legume family. Report SO-ITF-SM-49. New Orleans (LA): U.S.

Dept. of Agriculture Forest Service, Southern Forest Experiment Station. 5 p.

Parton WJ., and Risser PG. 1976. Osage site version of the ELM grassland model. In:

Proceedings of the 1976 summer computer simulation conference. La Jolla (CA):

Simulation Councils, Inc. p. 536-543.

Pellew RA. 1983. The impacts of elephant, giraffe, and fire upon the Acacia tortilis

woodlands of the Serengeti. African J Ecol. 21: 41-74.

Pennington TD, and Sarukhan J. 1968. Arboles tropicales de Mexico. Mexico City,

Mexico: Instituto Nacional de Investigaciones Forestales. 413 p.

Pieper RD. 1990. Overstory-understory relations in pinyon-juniper woodlands in New

Mexico. J Range Manage. 43: 413-415.

Polley HW, Johnson HB, and Mayeux HS. 1997. Leaf physiology, production, water use

and nitrogen dynamics of the grassland invader Acacia smallii at elevated CO2

concentrations. Tree Physiol. 17: 89-96.

Powell J, and Box TW. 1966. Brush management influences preference values of South

Texas woody species for deer and cattle. J Range Manage. 19(4): 212-214.

Texas Tech University, Pablo Teveni, August 2017

Powell J, Box TW, and Baker CV. 1972. Growth rate of sprouts after top removal of

huisache (Acacia farnesiana [L.] Willd.) (Leguminosae) in South Texas.

Southwest Nat. 17: 191-195.

Rasmussen GA, Scifres CJ, and Drawe DL. 1983. Huisache growth, browse quality, and

growth following burning. J Range Manage 36: 337-342.

Reinke S. 2007a. Ecological Site Characteristics: Blackland 24-44” PZ [Internet].

Washington (D.C.): U.S. Department of Agriculture, Soil Conservation Service.

[modified 2007 October 8; cited 2016 May 5]. Available from: https://esis.sc.

egov.usda.gov/ESDReport/fsReport.aspx?id=R150AY526TX&rptLevel=all&appr

oved=yes&repType=regular&scrns=&comm=

Reinke S. 2007b. Ecological Site Characteristics: Lakebed 25-35” PZ [Internet].

Washington (D.C.): U.S. Department of Agriculture, Soil Conservation Service.

[modified 2007 November 14; cited 2016 July 23]. Available from:

https://esis.sc.egov.usda.gov/ESDReport/fsReport.aspx?id=R150AY641TX&rptL

evel=all&approved=yes&repType=regular&scrns=&comm=

Reyes-Reyes G, Baron-Ocampo L, Cuali-Olivarez I, Frías-Hernández JT, Olalde-

Portugal V, Varela Fregoso L, and Dendooven L. 2002. C and N dynamics in soil

from the central highlands of Mexico as affected by mesquite (Prosopis spp.) and

huisache (Acacia tortuosa): A laboratory investigation. Appl Soil Ecol. 19: 27-34.

Reyes-Reyes G, Zamora-Villafranco E, Reyes-Reyes ML, Frías-Hernández JT, Olalde-

Portugal V, Dendooven L. 2003. Decomposition of leaves of huisache (Acacia

tortuosa) and mesquite (Prosopis spp.) in soil of the central highlands of Mexico.

Plant and Soil 256: 359-370.

Texas Tech University, Pablo Teveni, August 2017

Richardson CL. 1989. Brush management effects on deer habitat. Texas Agricultural

Extension Service. L-2347. 6 p.

Risser RG. 1981. The true prairie ecosystem. Stroudsburg (PA): Hutchinson Ross Publ.

Co. 557 pp.

Ruthven DC III, Braden AW, Knutson HJ, Gallagher JF, and Synatzske DR. 2003.

Woody vegetation response to various burning regimes in South Texas. J Range

Manage. 56(2):159-166.

Ruthven DC III, Fulbright TE, Beasom SL, and Hellgren EC. 1993. Long-term effects of

root plowing on vegetation in the eastern south Texas plains. J Range Manage.

46: 351-354.

Ruthven DC III, Hellgren EC, and Beasom SL. 1994. Effects of root-plowing on white-

tailed deer condition, population status, and diet. J Wildlife Manage. 58(1): 59-70.

San José JJ and Fariñas MR. 1991. Temporal changes in the structure of a Trachypogon

savanna protected for 25 years. Acta Oecol. 12: 237-47.

Sandford, WW, Usman S, Obot EO, Isichei AO, and Wari M. 1982. Relationship of

woody plants to herbaceous production in Nigerian Savanna. Trop Agr. 59: 315-

318.

Scanlan, JC, and Archer S. 1991. Simulated dynamics of succession in a North American

subtropical Prosopis savanna. J.Veg. Sci. 2(5): 625-634.

Schlesinger WH, Reynolds JE, Cunningham GL, Huenneke LF, Jarrell WM, Virginia

RA, and Whitford WG. 1990. Biological feedbacks in global desertification.

Science 247: 1043-1048.

Schmidly DJ. 2003. Wildlife legacy: changes during the 20th century. In: Forgason CA,

Texas Tech University, Pablo Teveni, August 2017

Bryant FC, and Genho PC, editors. Ranch Management: Integrating Cattle,

Wildlife, and Range: Proceedings of a Symposium. King Ranch Symposium: 150

Anniversary Celebration; 2003 October 1 and 2; Kingsville (TX). Kingsville

(TX): King Ranch. p. 59 to 73.

Scholes RJ and Archer SR. 1997. Tree-grass interactions in savannas. Ann Rev Ecol

Syst. 28: 517-544.

Scifres CJ. 1974. Salient aspects of huisache seed germination. Southwest Nat.18: 383-

392.

Scifres CJ. 1980. Brush Management: Principles and practices for Texas and the

Southwest. College Station: Texas A&M University Press. 360 p.

Scifres CJ, Mutz JL, Whitson RE, and Drawe DL. 1982. Interrelationships of huisache

canopy cover with range forage on the Coastal Prairie. J Range Manage. 35: 558-

562.

Smith HN and Rechenthin CA. 1964. Grassland restoration part 1: The Texas brush

problem. U.S. Department of Agriculture Soil Conservation Service Bulletin 4-

19114. 17 p.

Smith D, Paulsen GM, and Raguse CA. 1964. Comparative accuracy and efficiency in

determination of carbohydrates from grasses and legume tissue. Plant Physiol. 39:

960-962.

Sosebee RE. 1987. Water use by mesquite in West Texas. Lubbock: Texas Tech

University.

Sosebee RE. 2004. Timing: the key to successful brush and weed control with herbicides.

In: W.T. Hamilton, A. McGinty, D.N. Ueckert, C.W. Hanselka, and M.R. Lee,

100

Texas Tech University, Pablo Teveni, August 2017

eds. Brush management: Past, present, and future. College Station: Texas A&M

University Press. p. 85-96.

Sosebee RE and Dahl BE. 1991. Timing of herbicide application for effective weed

control: A plant’s ability to respond. Pp. 119-126 In: James LF, Evans JO, Ralphs

MH and Child RD, editors. Noxious range weeds. Boulder (CO): Westview Press.

p. 119-126.

Steuter, AA and Wright HA. 1980. White-tailed deer densities and brush cover on the

Rio Grande Plain. J Range Manage. 33(5): 328-331.

Stone CP, Smith CW, and Tunison JT, editors. 1992. Alien plant invasions in native

ecosystems of Hawai’i: Management and research. Honolulu: University of

Hawaii Press. 887 p.

Taylor RB, Rutledge J, Herrera JG. 1997. A field guide to common South Texas shrubs.

Austin: University of Texas Press. 106 p.

Teeri JA and Stowe LG. 1976. Climatic patterns and the distribution of C4 grasses in

North America. Oecologia. 23: 1-12.

Tiedemann AR and Klemmedson JO. 1986. Long-term effects of mesquite removal on

soil characteristics: I. Nutrients and bulk density. Soil Sci Soc Am J. 50: 472-75.

Tieszen LL. and Archer S. 1990. Isotopic assessment of vegetation changes in grassland

and woodland systems. In: Osmond CB, Pitelka LF, and Hidy GM. Plant biology

of the basin and range (Ecological studies 80). Berlin: Springer. p. 293-321.

Tischler, CR., Derner JD, Polley HW, and Johnson HB. 2004. Responses of seedlings of

five woody species to carbon dioxide enrichment. U.S. Department of Agriculture

Forest Service Proceedings. p. 161-163. RMRS-P-31.

101

Texas Tech University, Pablo Teveni, August 2017

Tutin, TG, editor. 1980. Flora Europea. Vol 5. London: Cambridge University Press.

U.S. Climate Data [Internet]. 2017. [cited 2017 April 10]. Available from:

http://www.usclimatedata.com/

U.S. Department of Agriculture, Natural Resources Conservation Service. 2017. Web

Soil Survey [Internet]. [modified 2016 August 10; cited 2017 April 14]. Available

from: https://websoilsurvey.sc.egov.usda.gov/

U.S. Department of Agriculture, Soil Conservation Service. 1985. Texas brush survey.

Temple, TX.

University of California at Davis. 2017. California Soil Resource Lab [Internet]. Davis:

University of California, Davis, Dept. of Land, Air, and Water Resources. [cited

on 2017 April 17]. Available from: https://casoilresource.lawr.ucdavis.edu/gmap/

Van Auken OW. 1994. Changes in competition between a C4 grass and a woody legume

with differential herbivory. Southwest Nat. 39(2): 114-121.

Van Auken OW and Bush JK. 1985. Secondary succession on terraces of the San

Antonio River. Bull Torrey Bot Club. 112(2): 158-166.

Van Auken OW and Bush JK. 1988. Competition between Schizachyrium scoparium and

Prosopis glandulosa. Am J Bot. 75(6): 782-789.

Van Auken OW and Bush JK. 1990. Interaction of two C3 and C4 grasses with seedlings

of Acacia smallii and Celtis laevigata. Southwest Nat. 35(3): 316-321.

Van Auken OW and Bush JK. 1991. Influence of shade and herbaceous competition on

the seedling growth of two woody species. Madroño 38: 149-157.

Virginia RA. 1986. Soil development under legume tree canopies. Forest Ecol Manage.

16: 69-79.

102

Texas Tech University, Pablo Teveni, August 2017

Virginia RA and Jarrell WM. 1983. Soil properties in a mesquite-dominated Sonoran

Desert ecosystem. Soil Sci Soc Am J. 53: 36-40.

Vora RS. 1989. Seed germination characteristics of selected native plants of the lower

Rio Grande Valley, Texas. J Range Manage. 42(1): 36-40.

Vora RS. 1990. Plant phenology in the Lower Rio Grande Valley, Texas. Texas Journal

of Science 42(2):

Walker BH, editor. 1987. Determinants of Tropical Savannas. Oxford (UK): IRL. 156 p.

Walker BJ and Noy-Meir I. 1982. Aspects of stability and resilience of savanna

ecosystems. In: Huntley BJ and Walker BH, editors. Ecology of tropical

savannas. Berlin: Springer-Verlag. p. 577-90.

Walker BH, Ludwig D, Holling CS, and Peterman RM. 1981. Stability of semi-arid

savanna grazing systems. J Ecol. 69: 473-98.

Walter H. 1972. Ecology of Tropical and Subtropical Vegetation. 2nd ed. Edinburgh

(UK): Oliver and Boyd. 688 p.

Wan C and Sosebee RE. 1990. Relationship of photosynthetic rate and edaphic factors to

root carbohydrate trends in honey mesquite. J Range Manage. 43: 171-176.

Welch TG. 1991. Is brush a problem? In: Welch TG, editor. Proceedings - Brush

Management Symposium; 1991 May 10-11; Giddings (TX). College Station

(TX): Texas Agricultural Extension Service. p. 1-5.

Welch TG. Undated. Optimum conditions for brush control. College Station: Texas

Agricultural Extension Service. p. 131-137.

Weltzin JF, Archer S, and Heitschmidt RK. 1997. Small-mammal regulation of

vegetation structure in a temperate savanna. Ecology 78(3): 751-763.

103

Texas Tech University, Pablo Teveni, August 2017

Weltzin, JF, and Coughenour MB. 1990. Savanna tree influence on understory vegetation

and soil nutrients in northwestern Kenya. J Veg Sci. 1: 325-34.

Whitson RE and Scifres CJ. 1980. Economic comparisons of alternatives for improving

honey mesquite-infested rangeland. Texas Agricultural Experiment Station

Bulletin 1307. 185 p.

Wright BD, Lyons RK, Cathey JC, and Cooper S. 2002. White-tailed deer browse

preferences for South Texas and the Edwards Plateau. College Station: Texas

A&M University Agrilife Extension. B6130

Wright HA and Bailey AW. 1982. Fire ecology: United States and southern Canada. New

York: John Wiley and Sons. 528 p.

Yeaton RI. 1988. Porcupines, fires and the dynamics of the tree layer of the Burkea

africana savanna. J of Ecol. 76: 1017-1029.

Xiangyun Z. 2002. Acacia farnesiana. In: Li ZY and Xie Y, editors. Invasive alien

species in China. Beijing: China Forestry Publishing House. p. 243.

Zisner CD. 1999. Seedlings identification and phenology of selected Sonoran Desert

dicotyledonous trees and shrubs. J Arizona-Nevada Acad Sci. 32(2): 129-154.

Zitzer SF, Archer SR, and Boutton TW. 1996. Spatial variability in the potential for

symbiotic N2 fixation by woody plants in a subtropical savannah ecosystem. J

Appl Ecol. 33: 1125-1136.

104

Texas Tech University, Pablo Teveni, August 2017

APPENDIX

Additional Figures

105

Texas Tech University, Pablo Teveni, August 2017

0.4 A 0.3

0.2 Content Average Daily Soil Water Content (m³/m³) 0.1

Volumetric Water Volumetric Permanent Wilting Point for clay at 0.3 m Field Capacity for clay at 0.3 m 0 0.4 B 0.3

0.2 Content

0.1 Volumetric Water Volumetric 0 C 0.4 0.3

0.2 Content

0.1 Volumetric Water Volumetric

0 D 0.5 0.4 0.3

0.2 Content 0.1

Volumetric Water Volumetric 0

Date

Figure A-1. Volumetric soil water content (vol./vol.) at 0.3 m for all study sites during period of study (mid-April 2012 to mid-November 2014), with field capacity and permanent wilting points for dominant soil types at 0.3 m: A = Baumgartner Ranch, B = Diebel Ranch, C = King Ranch, and D = Welder Wildlife Refuge.

106

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality 20

0 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 A Future-Delta-TNC (%)

100

Mortality (%) Mortality 20

0 40 60 80 100 120 140 160 180 B 4 Week Cumulative Precipitaion (mm)

Figure A-2. Best fit model for May data from all sites (A) Huisache Mortality vs. Future- Delta-TNC [next month TNC – current TNC] and B) Huisache Mortality vs. 4-week Cumulative Precipitation), according to AIC analysis (AIC = 67.804, AICc = 69.804, R2 = 0.577, Adj. R2 = 0.512, and n = 16) A) Function: y = -9.8546x + 93.768, R2 = 0.334 B) Function: y = -0.1819x + 112.15, R2 = 0.4299

107

Texas Tech University, Pablo Teveni, August 2017

100

Mortality (%) Mortality 20

0 0 20 40 60 80 100 120 140 160 180 200 4 Week Cumulative Precipitation (mm)

Figure A-3. Best fit model for October data from all sites (Huisache Mortality vs. 4- week Cumulative Precipitation), according to AIC analysis (AIC =83.022, AICc = 83.822, R2 = 0.07, Adj. R2 = 0.012, and n = 18) Function: y = -0.0648x + 101.35, R2 = 0.0701

100

40 Mortality (%) Mortality 20

0 0 2 4 6 8 10 12 14 16 Soil Available Water Content (%)

Figure A-4. Best fit model (1 of 2) for November data from all sites (Huisache Mortality vs. Soil AWC (available water content, %)) at 0.3 m, according to AIC analysis (AIC = 80.61, AICc = 81.533, R2 = 0.039, Adj. R2 = -0.029, and n = 16) Function: y = 0.504x + 88.315, R2 = 0.0394

108

Texas Tech University, Pablo Teveni, August 2017

100

40 Mortality (%) Mortality 20

0 15 16 17 18 19 20 21 22 Soil Temperature (°C) at 0.3 m

Figure A-5. Best fit model (2 of 2) for November data from all sites (Huisache Mortality vs. Soil AWC (available water content, %)) at 0.3 m, according to AIC analysis (AIC = 89.769, AICc = 90.569, R2 = 0.061, Adj. R2 = 0.003, and n = 18). Function: y = 1.6218x + 61.31, R2 = 0.0614

109