Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores

Rangeland Ecology & Management xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Rangeland Ecology & Management journal homepage: http://www.elsevier.com/locate/rama

Original Research

Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores☆ R.J. Ansley a,⁎, W.E. Pinchak b, M.K. Owens c a Professor and Department Head, Natural Resource Ecology and Management Department, Oklahoma State University, Stillwater, OK 74078, USA, (former Professor, Texas A&M AgriLife Research, Vernon, TX 76385, USA) b Professor, Texas A&M AgriLife Research, Vernon, TX 76385 c Professor and Associate Vice President, Oklahoma Agricultural Experiment Station, Stillwater, OK 74078, USA

a r t i c l e

i n f o

Article history: Received 15 January 2016 Received in revised form 28 January 2017 Accepted 30 January 2017 Available online xxxx Keywords: brush management cattle endozoochory Prosopis glandulosa seeds woody plant invasion

a b s t r a c t The dispersal of woody plant seeds by livestock has been implicated as one of the causes of woody plant encroachment in semiarid ecosystems worldwide. In the southern Great Plains, United States, cattle are suspected to have increased encroachment of the woody legume honey mesquite (Prosopis glandulosa Torr.) because they are effective consumers of mesquite pods and pass viable seed from those pods through their digestive systems. Since other animal species also consume or gather mesquite pods and seeds, our objective was to compare the removal of mesquite pods by cattle, other vertebrate herbivores, and insects. Mature pods were collected from trees in late summer and placed within each level of a hierarchical exclusion design using fences and cages that blocked cattle; other large vertebrates (deer, feral hogs); smaller vertebrates (rabbits, birds, rodents); and insects at replicate sites in north and south Texas locations. Pod removal was quantified during 60-d trials in the fall of each of 3 yr. The treatment that allowed cattle to have access to pods had the greatest or tied for the greatest pod removal at trial end in all trials. Final pod removal in the feral hog and white-tailed deer treatments was numerically lower but statistically similar (P ≤ 0.05) to cattle. However, the rate of pod removal during the first 20 d in several of the trials was greatest (P ≤ 0.05) in the cattle treatment at both locations. Pod removal by rodents was high in 1 yr at both locations, which we attributed to high growing season precipitation at both locations during that year. Results may have implications regarding seed-centric grazing management decisions and keeping cattle out of pastures when mesquite pods are abundantly present on the ground. © 2017 The Society for Range Management. Published by Elsevier Inc. All rights reserved.

Introduction In many arid and semiarid grasslands and savannas worldwide, the consumption and subsequent defecation of viable seeds of woody species by livestock (i.e., endozoochory) are partially responsible for the increased distribution and density of woody plants (Bahre and Shelton, 1993; Cox et al., 1993; D’Odorico et al., 2012; Lonsdale, 1993; Tews et al., 2004). In the southern Great Plains (SGP), United States, cattle and other domestic livestock may have played a role in increasing distribution of the woody indehiscent legume honey mesquite (Prosopis glandulosa Torr.) in the past 150 yr (Archer and Pyke, 1991; Brown and Archer, 1987, 1989). Because mesquite pods are indehiscent, relatively large (15− 20 cm long), and smooth textured, seed dispersal is dependent on endozoochory via pod consumption by large mammals, or the caching of individual seeds by rodents or insects, often after seeds have been deposited in large mammal feces (Duval et al., 2005;

☆ This study was funded by the US Department of Agriculture (USDA)−Cooperative State Research, Education, and Extension Service Grant 98-35315-6045 and USDA Hatch project 83107. ⁎ Correspondence: Jim Ansley, 008C Agricultural Hall, Oklahoma State University, Stillwater, OK 74078, USA. Tel.: +1-405-744-3014. E-mail addresses: [email protected], [email protected] (R.J. Ansley).

Weltzin et al., 1997). Mesquite pods (and seeds within) are too large to disperse by wind or water, or by attachment to animal fur or bird feathers. Mesquite pods are sweet to the taste and favored by cattle, regardless of forage grass availability (Glendening and Paulsen, 1955). Pod sugar content of mesquite and similar Prosopis species ranges from 27% to 32% (Del Valle et al., 1983; Felker, 1981; Marangoni and Alli, 1988). Passage through the cattle digestive system separates seeds from pods, scarifies the seed coat, and enhances germination of surviving seed (Campos and Ojeda, 1997; Peinetti et al., 1993). A greater percentage of seeds remain viable after passing through cattle than through sheep or goats (Kneuper et al., 2003). Mesquite seedlings readily establish from seeds that are in cattle dung pats (Brown and Archer, 1987; Kramp et al., 1998). A limited suite of native SGP herbivores also can pass viable mesquite seed through their digestive systems, including white-tailed deer (Odocoileus virginianus) and coyotes (Canis latrans) (Kramp et al., 1998). No data exist for feral hogs (Sus scrofa), a recent exotic invader in the SGP; however, Lynes and Campbell (2000) determined in Australia that viable Prosopis pallida seed passed through feral pigs. Most other native herbivores destroy mesquite seeds when they consume them, including lagomorphs (Bahre and Shelton, 1993) and rodents (Duval et al., 2005). Birds defecate viable seeds from many

http://dx.doi.org/10.1016/j.rama.2017.01.010 1550-7424/© 2017 The Society for Range Management. Published by Elsevier Inc. All rights reserved.

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

2

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

small-seeded rangeland species, including Opuntia and Juniperus (Ansley et al., 1995; Dean and Milton, 2000; García et al., 2010; Horncastle et al., 2004), but mesquite seeds are too large to survive passage through bird digestive systems. Insects, mainly bruchid beetles (Algarobius spp.) and conchuela (Chlorochora ligata Say), can destroy significant portions of Prosopis spp. seed crops (Kingsolver et al., 1977; Lerner and Peinetti, 1996; Smith and Ueckert, 1974; Van Klinken and White, 2011; Zimmermann, 1991) mostly after pods have fallen to the ground (Watts et al., 1989). Several studies have addressed mesquite seed viability, germination, or establishment after passage through cattle or other animals (Bush and Van Auken, 1990; Campos et al., 2011; Kneuper et al., 2003; Kramp et al., 1998; Peinetti et al., 1993). However, few studies have compared the rate of pod consumption (or removal) among coexisting animal groups in a natural setting (Villagra et al., 2002; Weltzin et al., 1997), and no studies that we know of have compared cattle with native herbivores. It likely would be easier for a large-mandible species like cattle than it would for smaller herbivores to rapidly consume whole pods. Janzen (1982) found in Costa Rica that cattle were very effective at consuming large indehiscent guanacaste (Enterolobium cyclocarpum) pods. We have observed cattle rapidly consuming clusters of mesquite pods on the ground (Ansley, unquantified observation). In addition, the larger energy requirements of cattle and greater body mass per unit land area, which is a function of body mass × stocking rate, could cause more rapid pod consumption by cattle than by native herbivores in any given area. Somewhat contrary to this hypothesis, Kneuper et al. (2003) found that the presence or absence of livestock (cattle, sheep, or goats) did not affect disappearance rates of mesquite seed pods in one pasture when compared with another pasture that contained only native herbivores (mainly white-tailed deer). The animal guilds that now exist within mesquite savannas in the SGP are different than when Europeans first settled the area. There are now more deer (Wolverton et al., 2007), and feral hogs were not present until the 1980s. However, a better understanding of consumption of pods by cattle and other animal groups would offer more insight regarding the factors that may have led to mesquite expansion in the SGP and could potentially affect livestock management decisions today. Our objective was to compare the removal of mesquite pods by cattle, other vertebrate herbivores, and insects.

At the Uvalde replicates, the 30-yr mean annual precipitation is 617 mm. Average annual maximum and minimum temperatures were 35.5°C and 13.7°C, respectively. Growing season can range from 250 to 365 d. Uvalde 1 has nearly level to gently sloping, deep, calcareous upland soils of the Knippa Clay series. Uvalde 2 has nearly level to gently sloping, deep, calcareous alluvial soils of the Uvalde silty clay loam series. Vegetation at both Uvalde sites consisted of a mixed thorn shrub community containing honey mesquite and liveoak (Quercus virginiana) trees, shrubs guajillo (Acacia berlandieri), blackbrush (Acacia rigidula), whitebrush (Aloysia gratissima) and pricklypear cactus (Opuntia lindheimeri), and grasses red grama (Bouteloua trifida), Wright’s threeawn (Aristida purpurea), common curly-mesquite (Hilaria belangeri), buffalograss, and Texas wintergrass. A diverse forb cover varies with rainfall pattern and abundance. Grass production was higher at the Vernon replicates (~3 000 kg· ha−1) than the Uvalde replicates (~1 200 kg· ha−1). Domestic cattle (Bos taurus) were grazed as cow-calf operations at all replicates; stocking rates were considered “moderate” with one animal unit (AU) to 10 ha at Vernon 1, one AU to 12 ha at Vernon 2 and 3, and one AU to 35 ha at Uvalde 1 and 2. Each replicate was visited by a different herd of cattle. No native herbivore population surveys were collected in this study. However, species that have been frequently observed at all 5 replicates were white-tailed deer, feral hogs, javelina, coyotes, jackrabbits, cottontail rabbits, raccoons, skunks, and numerous bird, rodent, and insect species. In addition, fresh dung pats and fecal pellets of cattle, deer, and hogs were observed at each of the sites. All replicates were N 20 km from one another, so it was probable that each replicate was visited by different individuals of each species, although this was not verified. Each replicate contained a hierarchical series of fence and/or wire mesh barriers to progressively allow additional animal group access to mesquite pods (Fig. 1). Each exclusion level was referred to as a “treatment” and was numbered and labeled according to the next animal group (plus all previous animal groups) that was allowed access to mesquite pods (Table 1). The order of treatments progressed generally from smaller to larger animals and included 1) None—all animals excluded; 2) insects; 3) rodents, 4) birds, 5) rabbits; 6) white-tailed deer and raccoons; 7) feral hogs, javelin, and coyotes; and 8) cattle. The first three levels included tops on the cages or fences to prevent bird access.

Methods The study was conducted on private ranches at two locations 520 km apart: near Vernon in north central Texas and near Uvalde in south Texas (hereafter Vernon and Uvalde are referred to as “locations”). Three 3-ha replicate sites were established near Vernon (Vernon 1, Smith Walker Ranch 34°01′N, 99°14′W, elevation 374 m; Vernon 2, Peach Orchard Pasture 33°55′N, 99°04′W, elevation 340 m; Vernon 3, Ninemile Pasture 33°51′N, 99°25′W, elevation 382 m), and two replicate sites (hereafter “replicates”) were located near Uvalde (Uvalde 1, Harris Ranch 29°19′N; 100°05′W; elevation 333 m; Uvalde 2, Turkey Creek Ranch, 29°04′N; 100°01′W; elevation 253 m). All five replicates were randomly located within mesquite-dominated rangeland communities. At the Vernon replicates, the 30-yr mean annual precipitation is 653 mm and mean annual air temperature is 16.9°C, with the peak in July (29.2°C) and low in January (3.8°C) (NOAA-NCDC, 1997). Growing season is typically from early March through October (~240 d). Soils at each Vernon site were fine, mixed, superactive, thermic Vertic Paleustolls of the Tillman clay loam series with 0 − 1% slopes (USDANRCS, 2014a), and vegetation consisted of a dominant mesquite overstory, lightly scattered lotebush (Ziziphus obtusifolia [T.&G.] Gray) shrubs, and grasses Texas wintergrass (Nassella leucotricha [Trin. & Rupr.] Pohl), buffalograss (Bouteloua dactyloides [Nutt.] J.T. Columbus), vine mesquite (Panicum obtusum Kunth), and sand dropseed (Sporobolus cryptandrus [Torr.] A. Gray) (USDA-NRCS, 2014b).

Figure 1. Treatment exclosure design for each replicate. Numbers correspond to the numbers in the left hand column of Table 1. A detail of the pod arrangement for levels 4 through 8 is shown on the left. Dashed line around squares 5-8 indicates either unfenced areas within a larger fence (5-7) or an unfenced area (8). For each area, 16 groups of 20 pods each were distributed in a 4 by 4 grid. In treatment 3 (bird), the 16 groups were divided into two groups of 8 each. For treatments 1 and 2 (insect and rodent cages) each cage included one 20-pod group randomly located throughout the deer or hog enclosures. They are shown here more clustered together than they actually were.

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

3

Table 1 Design specifics for each animal access level Treatment ID1

Animal groups with access

Exclosure description and (in parentheses) no. of exclosure units within each replicate

1. None

None

2. Insect

I

3. Rodent

I, Ro

4. Bird

I, Ro, B

5. Rabbit 6. Deer 7. Hog 8. Cattle

I, Ro, B, Ra I, Ro, B, Ra, D I, Ro, B, Ra, D, H I, Ro, B, Ra, D, H, C

0.3 m diameter × 0.3 m tall cylinders wrapped with 1-mm metal mesh screen with metal bottom, set 5 cm into soil with soil covering metal bottom (16; 8 each in hog and deer exclosures). 0.5 m diameter × 0.5 m tall cylinders wrapped with 1.3-cm galvanized metal mesh hardware cloth; open bottom (16; 8 each in hog and deer exclosures). 5 × 5 m × 1.2 m tall combination panel with poultry net covering sides and top; bottom 5 cm on sides left open for rodent access; within the cattle exclosure but not in deer or hog exclosure (2). 10 × 10 m × 1.2 m tall combination panel with poultry net wrap from ground to third course; open top; within the deer exclosure (1). 81 × 46 m × 3 m tall deer fence in different area than hog fence within the cattle exclosure (1). 46 × 46 m × 1.2 m tall hog panel within the cattle exclosure (1). 127 × 127 m × 1.2 m tall 5 strand barbed-wire fence (1). Open rangeland outside exclosure complex (1).

I, insects; Ro, rodents; B, birds; Ra, rabbits; D, deer; H, feral hogs; C, cattle. 1 Treatment ID from 1 to 8 identifies the next animal group with access plus whatever previous groups that have access.

Treatments 4 through 7 were open topped with variations of type of perimeter fencing. Treatments 6 and 7 were referred to as “Deer” and “Hogs,” respectively, because we assumed those were the main animal species that would consume mesquite pods. However, other animal species listed in those treatments would also have access, although their population densities are typically much lower than deer or feral hogs on these sites. Treatment 8, referred to as “Cattle” because this was the only treatment where cattle had access to pods, was open to all animals. Two replicates were operational in 1999, and three were operational in 2000 and 2001 at Vernon. Two replicates were operational at Uvalde in 2000 and 2001. Mesquite pod removal trials were initiated for each replicate toward the end of the mesquite growing season (late July or August in Uvalde; September in Vernon) shortly after normal pod drop occurred. Currentyear mesquite pods were collected from nearby trees a few weeks before each trial and were placed in groups of 20 undamaged pods within a 0.25-m 2 area on the ground either within small screened cages (treatments 1 and 2) or in a particular exclosure (treatments 3−7), or outside any exclosure (treatment 8). Sixteen groups of 20 pods each were included for each treatment such that a total of 2 560 pods were distributed per replicate per trial (20 pods per group × 16 groups per treatment × 8 treatments). Each pod was 15 − 20 cm long and contained about 20−25 seeds. Following pod placement, pod removal from each 20-pod group was measured at 7, 14, 20, 30, 40, and 60 d post placement (at Uvalde the first two measurements were at 4 − 5 and 10 − 11 d) and quantified as a percentage of original number of pods removed. A total of 12 pod removal trials were completed during the study: 2 replicates at Vernon in 1999, 3 reps at Vernon and 2 reps at Uvalde in 2000, and 3 reps at Vernon and 2 reps at Uvalde in 2001. During all trials each replicate had a background presence of mesquite pods that had fallen from trees. Thus, the pods used in each trial blended in with the background level of pods.

because we did not have data from Uvalde in that year. The model used location by treatment to test the main effects of location and treatment and year by treatment to test the main effects of year. Treatmentby-year and location-by-year interactions were tested using the pooled error. Separate analyses were conducted for the two common trial sample periods—20 and 60 d. A third analysis assessed differences in rate of harvest among animal treatment levels during the trials. A repeated measures analysis was performed using PROC GLM with animal exclusion treatment (eight levels) and sample period (six levels) as main effects and percent pod removal as the response variable. Because of slight differences in sampling schedules during the trials, these analyses were done within each location and year. Before all analyses, percent pod removal data were transformed using Arcsine transformation (Zar, 1996). Means were compared using least significant difference at P ≤ 0.05. F values reported in the tables were from transformed values; however, percent pod removal means reported in the figures are actual means. Results The first analysis found a significant (P ≤ 0.0001) main effect of animal exclusion treatment at both 20 and 60 d when pooled over both locations and all 3 yr. Percent pod removal in treatment 8 was significantly (P ≤ 0.05) greater than all other treatments at 20 d but was similar to treatments 6 and 7 at 60 d (Fig. 2). Significant (P ≤

Analysis We initially tested effects of animal treatment on percent pod removal at two time periods during the pod removal trials, 20 and 60 d, that were common to all replicates at both locations and in all 3 yr. We used PROC GLM with animal treatment (8 levels) as the main effect and percent pod removal as the response variable. In this analysis, n = 12, with 2 replicates at Vernon in 1999, 3 replicates at Vernon in 2000 and 2001, and 2 replicates at Uvalde in 2000 and 2001. A second analysis tested the main effects of location, year, and animal treatment and their interaction on pod removal. A repeated measures analysis was performed using PROC GLM (SAS, 2002) with location (Vernon, Uvalde); animal exclusion treatment (8 levels); and yr (2000, 2001) as main effects. Data from 1999 were not included

Figure 2. Means of mesquite percent pod removal for each animal access treatment at 20 and 60-day sample periods during pod removal trials when pooled over all locations and years (n=12). Vertical lines above bars are standard error. Means with similar letters within each sample period are not significantly different at P ≤ 0.05. Asterisk indicates significant (P ≤ 0.05) difference between days within an animal treatment.

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

4

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

Table 2 F values of GLM analyses testing main effects of Location (Vernon, Uvalde), Treatment (8 levels), and Yr (2000, 2001) and their interactions at 2 time periods, 20 and 60 d, during pod removal trials for the response variable percent pod removal. Vernon had 3 replicates and Uvalde had 2 replicates. Bottom rows indicate tests of Location and Treatment using the Location × Treatment error term, and test of Yr using the Treatment × Yr error term. All analyses were based on transformed values of percent pod removal. Asterisk indicates significance at P ≤ 0.05. Source

df

Location Treatment Location × Treatment Yr Treatment × Yr Location × Yr Error Location Treatment Location × Treatment Yr Treatment × Yr

1 7 7 1 7 1 55 1 7 7 1 7

Pr N F D 20

D 60

b 0.0001* b 0.0001* 0.046* 0.012* 0.353 0.044*

0.097 b 0.0001* 0.128 b 0.0001* 0.107 0.566

0.015* 0.014*

0.236 0.001*

0.046*

0.009*

0.05) differences between the two sample periods occurred with treatments 5, 6, and 7. The second analysis showed significant (P ≤ 0.05) main effects of location, treatment, and year at 20 d, and significant main effects of treatment and year at 60 d during the pod removal trials (Table 2). There were significant location × treatment and location × year interactions at 20 d but no significant interactions at 60 d. Thus, the relationships among animal exclusion treatments changed during the course of the trials. When pooled over all treatments and both years, pod removal at 20 d was significantly (P ≤ 0.05) greater at Vernon (41.8% ± 6.1) than Uvalde (14.9% ± 4.2); by 60 d there was no difference between locations (Vernon 55.1% ± 6.3; Uvalde 46.8% ± 6.2). Pod removal was significantly higher in 2000 than 2001 at 20 d (39.0 % ± 6.7 vs. 23.1% ± 5.1) and 60 d (62.3% ± 6.7 vs. 41.3% ± 5.5). The third analyses that tested effects of animal treatment on each sample day found significant (P ≤ 0.05) main effects of treatment and days for both locations and each year and a significant treatment × days interaction in all 3 yr at Vernon and in 2001 at Uvalde (Table 3). At Vernon, the rapid removal of N 50% of pods within the first 3 wk was detected in several treatments in 2000 and 2001 but not in 1999 (Fig. 3). In 2001, pod removal was initially highest in treatments 8 (cattle) and 7 (hogs) (Fig. 3C). Pod removal in treatment 6 (deer) was initially slower than treatments 7 and 8 but increased over time to levels similar to those treatments. A similar trend was seen in 2000 (Fig. 3B). Pod removal in treatment 5 (rabbits) had a delayed response in 2000

Table 3 F values of GLM analyses testing main effects of Treatment (“Trt”; 8 levels) and days during pod removal trials (“D”; 6 levels) and the Treatment × Days interaction for the response variable percent pod removal for each location and year. Vernon 1999 and Uvalde 2000 and 2001 had 2 replicates, and Vernon 2000 and 2001 had 3 replicates per treatment. Bottom row indicates test of Treatment using Rep × Treatment as the error term. Effects of Days and Treatment × Days were tested using the pooled error. All analyses were based on transformed values of percent pod removal. Source

Rep Trt Rep × Trt D Trt × D Error Trt Rep × Trt

df

1 7 7 5 35 40 7 7

Pr N F

df

Vernon 1999

Uvalde 2000

Uvalde 2001

0.009 b 0.0001 b 0.0001 b 0.0001 b 0.0001

0.042 b 0.0001 0.0002 b 0.0001 0.354

0.0009 b 0.0001 b 0.0001 b 0.0001 0.013

0.015

0.029

0.034

2 7 14 4 28 64 7 14

Pr N F Vernon 2000

Vernon 2001

b 0.0001 b 0.0001 b 0.0001 b 0.0001 0.005

0.021 b 0.0001 b 0.0001 b 0.0001 b 0.0001

0.0001

b 0.0001

Figure 3. Mesquite pod removal for each animal access treatment on different days during removal trials at Vernon, 1999-2001. Vertical lines are standard error bars. Means with similar letters at each sample period within each panel are not significantly different at P ≤ 0.05.

that was similar to treatment 6 (see Fig. 3B). Regarding treatment 3 (rodents), almost all of the final pod removal value of 53.7% in 2000 occurred by d 20 (see Fig. 3B). The 53.7% value in 2000 was much higher than in 1999 (9.0%) or 2001 (20.3%). At Uvalde, pod removal was initially more rapid in treatment 8 (cattle) compared with other treatments in both years (Fig. 4). Similar to Vernon, pod removal in treatment 6 (deer) started slowly but increased over time to levels similar to treatment 8 in the yr 2000, and to a lesser degree in 2001. Pod removal in treatment 7 (hog) was not as rapid as at Vernon in either year and, in 2000, exhibited a delayed response similar to that of treatment 6 (Fig. 4A). Pod removal in the treatment 3 (rodents) finished at 56.8% in 2000, similar to Vernon, but the pattern of removal was different as most occurred in the last 20 d instead of the first 20 d, as was found at Vernon (see Figs. 3B and 4A). The 56.8% in 2000 was much higher than in 2001 (13.8%). Final percent pod removal at trial end (d 60) is shown in bar chart form in Figure 5. Pod removal in treatments 1 (no access) and 2 (insects only) was low in all trials. Slight increases in both these treatments at Uvalde were observed to be due to rodents gaining access to pods within some of the insect and rodent cages rather than this being attributed to insect activity. Final percent pod removal tended to be greatest in treatments 6, 7, and 8 and was numerically greatest or tied for greatest in treatment 8, but it was never significantly (P ≤ 0.05) greater in treatment 8 over all other treatments. However, as Figures 3 and 4 revealed, at earlier times during some of the trials, pod removal in treatment 8

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

5

Discussion

Figure 4. Mesquite pod removal for each animal access treatment on different days during removal trials at Uvalde, 2000-2001. Vertical lines are standard error bars. Means with similar letters at each sample period within each panel are not significantly different at P ≤ 0.05. Means without letters are bracketed between letters that apply to them and have been omitted to improve visual clarity.

was significantly (P ≤ 0.05) greater than all of the other 7 treatments (see Fig. 3C, d 20 and d 30; Fig. 4A and B, d 10, d 20, and d 30). This early harvest effect in treatment 8 was more evident at Uvalde than Vernon.

The study revealed that there was a location effect in which animal guilds in the north location (Vernon) were generally more aggressive harvesting pods early in the trials, but this difference disappeared by trial’s end. There also was a yearly effect where pod harvest was greater at both locations in the yr 2000 compared with 2001. Regarding comparison of responses among animal exclusion treatments, within the hierarchical treatment design we employed, one cannot say definitively that within each new treatment level the new animal group with pod access is solely responsible for any increases in pod removal. This is because each treatment level includes pod removal by the new animal group in addition to all other groups that previously had access to pods. However, while we cannot be certain of a direct association between each treatment level and the animal group targeted for that level, the consistency in incremental increases in pod removal in response to increasing additions of animal groups that we observed over locations and years reinforces the assumption of a link between pod removal and the targeted animal group in each treatment level. Among the treatments, there was a major division of pod access that occurred between treatments 5 (rabbit) and 6 (deer). Animals with access from treatments 2 through 5 were too small to be impeded by the cattle fence that separated treatment 7 from treatment 8. Thus, pod removal in treatment 5 likely represents the maximum extent of pod harvest by small animals. This assumption is supported by the fact that in every trial, final percent pod removal was as great or greater in treatment 5 than in treatments 2, 3, or 4. Because of this, we can also conclude that any pod removal in treatments 6, 7, or 8 that exceeded that found in treatment 5 likely represented the effects of deer, hogs, or cattle. Among treatments 1 through 5, one factor that stands out is the apparent effect of rodents (treatment 3) on pod removal at both locations in the yr 2000. This response may have been related to greater April−August precipitation in 2000 than 2001 (307 vs. 144 mm at Vernon; 323 vs. 175 mm at Uvalde), which would have generated more grass growth and probably increased rodent populations. It may also have related to the background amount of mesquite pods, which was

Figure 5. Means of mesquite pod removal percent for each animal access treatment during 3 years at Vernon and 2 years at Uvalde. Vertical lines above bars are standard error. Means with similar letters within each panel are not significantly different at P ≤ 0.05. Treatment number on X-axis follows that in Figure 2 and Table 1 (1-None, 2-Insects, 3-Rodents, 4-Birds, 5-Rabbits, 6-Deer, 7-Hogs, and 8-Cattle).

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

6

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

likely greater in 2001 than 2000 since pod production is usually higher in drier years (Felker et al., 1984; Lee and Felker, 1992). More background pods plus fewer rodents in 2001 may have diluted harvest opportunities in our targeted pod clusters. Regarding effects of large mammals in treatments 6, 7, and 8, the final percent pod removal was greatest or tied for greatest in treatment 8, although it was never significantly (P ≤ 0.05) greater in this treatment over all other treatments. However, the rate of pod removal was significantly (P ≤ 0.05) greater in treatment 8 than the other treatments at both locations in certain years. These results suggest that cattle had an effect, especially during the first few weeks of the trials. Alternatively, it is possible that deer or hogs instead of cattle caused the increased pod removal in treatment 8 by first foraging for pods outside the cattle fence, then crossing over the fence and foraging for pods inside the fenced area. However, we note that each replicate plot was large enough so that at any point along the barbed-wire cattle fence, deer and feral hogs would have perceived it as an ordinary pasture fence that they could easily jump over (deer) or pass under (hogs) (Ansley, personal observations). Thus, we are reasonably confident that deer and hogs readily foraged on both sides of the cattle fence and that the extra pod removal outside the fence was due to cattle. Mesquite Recruitment Ecology and Management Figure 6 illustrates where the essential step of mesquite pod removal by animals fits within the broader context of mesquite recruitment ecology and mesquite management. Various abiotic factors affect pod production and, once pods containing seeds drop to the ground, they are either consumed, gathered, or destroyed or become part of the soil

seed bank. Cattle, deer, hogs, and many other wildlife species may contribute to this survival success by endozoochory or by caching pods and/ or seeds. In contrast, insects, birds, and some smaller mammals more likely destroy the seeds. A variety of human management inputs can arrest this process at many points in the life cycle. Most strategies involve costly treatments that kill mature trees, which has the dual effect of decreasing pod production and increasing the competitive effects of grasses on mesquite seedling emergence from the existing mesquite seed bank (Van Auken and Bush, 1997). In addition, disturbances such as fire may also kill emerging seedlings (Ansley et al., 2015). Grazing management practices that prevent cattle from grazing in pastures during times when viable mesquite pods are on the ground may also be an option, but this has not been widely implemented. Others have noted that management actions that limit consumption of seeds of unwanted woody species by livestock may be a key component of an ecologically sustainable brush management strategy (Grice, 1998; Lonsdale, 1993; Lonsdale et al., 1988; Steenkamp and Chown, 1996; Zimmermann, 1991). Mesquite Pod Predation During SGP Settlement Our study suggests that cattle may compete for mesquite pods in the presence of modern wildlife species guilds, including new species such as feral hogs. The relative influence of a large mammal-like cattle on mesquite pod removal may actually have been greater in the late 1800s, when cattle were introduced to the SGP, than it is today. Deer populations were lower than they are today (Wolverton et al., 2007), prairie dog populations were higher (Weltzin et al., 1997), and feral hogs were not present in the SGP. Bison (Bison bison) were present in

Figure 6. Flow chart depicting processes related to mesquite seedling establishment. Black solid lines indicate major processes; black dashed lines indicate probable processes. Gray dashed lines indicate human management activities that affect the process at different entry points. Gray-filled boxes are the focal points of the current study. “Cattle Harvest” and “Wildlife Harvest” boxes represent endozoochory (cattle, large native herbivores) or being cached by certain wildlife species such as rodents. In contrast, insects would have a more predatory role in destroying pods and seeds as they consume portions of them.

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

the SGP before European settlement (Dillehay, 1974; Hornaday, 1889; Huebner, 1991), and there is evidence of bison passing viable grass seeds (Rosas et al., 2008), but for reasons not understood, bison did not cause accelerated mesquite encroachment as happened after cattle arrived. Other anthropogenic-related factors, including reduced fire frequency and reduced competition from grasses, both due to livestock overgrazing, may also have contributed to the accelerated mesquite distribution in the SGP (Van Auken, 2000). However, studies on fire effects on mesquite seedlings (Ansley et al., 2015; Wright et al., 1976) show only a 2-yr window of vulnerability, and mesquite seedlings were found to establish in dense stands of grass (Ansley et al., 2015; Brown and Archer, 1999). These studies imply that seed distribution via endozoochory was an important factor in accelerating mesquite invasion and that cattle were more successful than bison in distributing viable mesquite seed. No studies have directly measured expansion of mesquite plants via cattle endozoochory in North America. However, we suggest that it was the introduction of cattle in high numbers, which could retain seed in their digestive system for up to 8 d (Gardener et al., 1993; Hogan and Phillips, 2011; Janzen, 1982; Traveset, 1998), that provided the first long-distance mesquite seed disperser in the region, especially if large herds were moved by ranchers from one location to another. Several studies point to the native herbivore population of deer (Brown and Archer (1987), rodents (Campos et al., 2007; Cox et al., 1993; Duval et al., 2005; Valone and Thornhill, 2001), and insects (Villagra et al., 2002; Weltzin et al. 1993) as being only short-distance dispersers of mesquite seed. Management Implications Results from this study show that, while both cattle and native herbivores contributed to the consumption or removal of mesquite pods (and seeds), cattle may have consumed pods at a faster rate than other animal groups if, in fact, treatment 8 responses were due to cattle. Grazing strategies that exclude cattle from pastures when mesquite pods are most palatable and abundantly available on the ground may be an option to help reduce recruitment opportunities for this woody species. The development of alternate pastures, perhaps with introduced grasses, as “pod-free” pastures where cattle can be contained during periods of high pod availability in the mesquite-dominated pastures, may facilitate this process. Weathering effects and damage to mesquite pods and seeds from insects may limit the length of time that mesquite seeds remain viable on the ground so a relatively short period of avoidance of pod-rich pastures by cattle could be accomplished. In the SGP, the avoidance period would typically need to be 6−7 mo between pod drop in August to March of the next year. A better understanding of the timing and ecology of weathering and insect effects on pods and seeds will help refine the timing of such seed-centric cattle grazing strategies. In addition to grazing management, other treatments may be used, such as prescribed fire in the late-summer and fall months to destroy pods that have dropped to the ground. Acknowledgments We thank Matt Angerer, Rose Cooper, Doug Fulford, David Jones, Betty Kramp, and Doug Tolleson for assisting with construction of the field sites and pod removal evaluations. We also thank the Smith and Walker families and the W. T. Waggoner Estate for providing land area for this study. References Ansley, R.J., Kramp, B.A., Jones, D.L., 2015. Honey mesquite (Prosopis glandulosa) seedling responses to seasonal timing of fire and fire intensity. Rangeland Ecology & Management 68, 194–203.

7

Ansley, R.J., Pinchak, W.E., Ueckert, D.N., 1995. Changes in redberry juniper distribution in northwest Texas (1948 to 1982). Rangelands 17, 49–53. Archer, S., Pyke, D.A., 1991. Plant-animal interactions affecting plant establishment and persistence on revegetated rangeland. Journal of Range Management 44, 558–565. Bahre, C.J., Shelton, M.L., 1993. Historic vegetation change, mesquite increases and climate in southwestern Arizona. Journal of Biogeography 20, 489–504. Brown, J.R., Archer, S., 1987. Woody plant seed dispersal and gap formation in American subtropical savanna woodland: the role of domestic herbivores. Vegetatio 73, 73–80. Brown, J.R., Archer, S., 1989. Woody plant invasion of grassland: establishment of honey mesquite (Prosopis glandulosa var. glandulosa) on sites differing in herbaceous biomass and grazing history. Oecologia 80, 19–26. Brown, J.R., Archer, S., 1999. Shrub invasion of grassland: recruitment is continuous and not regulated by herbaceous biomass or density. Ecology 80, 2385–2396. Bush, J.K., Van Auken, O.W., 1990. Growth and survival of Prosopis glandulosa seedlings associated with shade and herbaceous competition. Botanical Gazette 151, 234–239. Campos, C.M., Campos, V.E., Mongeaud, A., Borghi, C.E., De Los Rios, C., Giannoni, S.M., 2011. Relationships between Prosopis flexuosa (Fabaceae) and cattle in the Monte desert: seeds, seedlings and saplings on cattle-use site classes. Revista Chilena de Historia Natural 84, 289–299. Campos, C.M., Giannoni, S.M., Taraborelli, P., Borghi, C.E., 2007. Removal of mesquite seeds by small rodents in the Monte desert, Argentina. Journal of Arid Environments 69, 228–236. Campos, C.M., Ojeda, R.A., 1997. Dispersal and germination of Prosopis flexuosa (Fabaceae) seeds by desert mammals in Argentina. Journal of Arid Environments 35, 707–714. Cox, J.R., De Alba-Avila, A., Rice, R.W., Cox, J.N., 1993. Biological and physical factors influencing Acacia constricta and Prosopis velutina establishment in the Sonoran Desert. Journal of Range Management 46, 43–48. Dean, W.R.J., Milton, S.J., 2000. Directed dispersal of Opuntia species in the Karoo, South Africa: are crows the responsible agents? Journal of Arid Environments 45, 305–314. Del Valle, F.R., Escobedo, M., Munoz, M.J., Ortega, R., Bourges, H., 1983. Chemical and nutritional studies on mesquite beans (Prosopis juliflora). Journal of Food Science 48, 914–919. Dillehay, T.D., 1974. Late quaternary bison population changes on the southern plains. Plains Anthropologist 19, 180–196. D’Odorico, P., Okin, G.S., Bestelmeyer, B.T., 2012. A synthetic review of feedbacks and drivers of shrub encroachment in arid grasslands. Ecohydrology 5, 520–530. Duval, B.D., Jeckson, E., Whitford, W.G., 2005. Mesquite (Prosopis glandulosa) germination and survival in black-grama (Bouteloua eriopoda) grassland: relations between microsite and heteromyid rodent (Dipodomys spp.) impact. Journal of Arid Environments 62, 541–554. Felker, P., 1981. Uses of tree legumes in semiarid regions. Economic Botany 35, 174–186. Felker, P., Clarke, P.R., Osborn, J.F., Cannell, G.H., 1984. Prosopis pod production—comparison of North American, South American, Hawaiian, and African germplasm in young plantations. Economic Botany 38, 36–51. García, D., Zamora, R., Amico, G.C., 2010. Birds as suppliers of seed dispersal in temperate ecosystems: conservation guidelines from real-world landscapes. Conservation Biology 24, 1070–1079. Gardener, C.J., McIvor, J.G., Jansen, A., 1993. Passage of legume and grass seeds through the digestive tract of cattle and their survival in feces. Journal of Applied Ecology 30, 63–74. Glendening, G.E., Paulsen, H.A., 1955. Reproduction and establishment of velvet mesquite as related to invasion of semidesert grasslands. US Department of Agriculture Forest Service Technical Bulletin 1127, Washington, DC, USA 50 pp. Grice, A.C., 1998. Ecology in the management of Indian jujube (Ziziphus mauritiana). Weed Science 46, 467–474. Hogan, J.P., Phillips, C.J.C., 2011. Transmission of weed seed by livestock: a review. Animal Production Science 51, 391–398. Hornaday, W.L., 1889. The extermination of the American bison. Smithsonian Institution Annual Report, 1887 2, pp. 367–548. Horncastle, V.J., Hellgren, E.C., Mayer, P.M., Engle, D.M., Leslie Jr., D.M., 2004. Differential consumption of Eastern Red Cedar (Juniperus virginiana) by avian and mammalian guilds: implications for tree invasion. American Midland Naturalist 152, 255–267. Huebner, J.A., 1991. Late prehistoric Bison populations in central and southern Texas. Plains Anthropologist 36, 343–358. Janzen, D.H., 1982. Differential seed survival and passage rates in cows and horses, surrogate Pleistocene dispersal agents. Oikos 38, 150–156. Kingsolver, J.M., Johnson, C.D., Swier, S.R., Teran, A., 1977. Prosopis fruits as a resource for invertebrates. In: Simpson, B.B. (Ed.), Mesquite: its biology in two desert scrub ecosystems, US/IBP Synthesis Series No. 4. Dowdon, Hutchinson & Ross, Inc, Stroudsburg, PA, USA, pp. 108–122. Kneuper, C.L., Scott, C.B., Pinchak, W.E., 2003. Consumption and dispersion of mesquite seeds by ruminants. Journal of Range Management 56, 255–259. Kramp, B.A., Ansley, R.J., Tunnell, T.R., 1998. Survival of mesquite seedlings emerging from cattle and wildlife feces in a semi-arid grassland. The Southwestern Naturalist 43, 300–312. Lee, S.G., Felker, P., 1992. Influence of water/heat stress on flowering and fruiting of mesquite (Prosopis glandulosa var. glandulosa). Journal of Arid Environments 23, 309–319. Lerner, P., Peinetti, R., 1996. Importance of predation and germination on losses from the seed bank of Calden (Prosopis caldenia). Journal of Range Management 49, 147–150. Lonsdale, W.M., 1993. Rates of spread of an invading species—Mimosa pigra in northern Australia. Journal of Ecology 81, 513–521. Lonsdale, W.M., Harley, K.L.S., Gillett, J.D., 1988. Seed bank dynamics in Mimosa pigra, an invasive tropical shrub. Journal of Applied Ecology 25, 963–976. Lynes, B.C., Campbell, S.D., 2000. Germination and viability of mesquite (Prosopis pallida) seed following ingestion and excretion by feral pigs. Tropical Grasslands 34, 125–128.

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010

8

RJ. Ansley et al. / Rangeland Ecology & Management xxx (2017) xxx–xxx

Marangoni, A., Alli, I., 1988. Composition and properties of seeds and pods of the tree legume Prosopis juliflora (DC). Journal of the Science of Food and Agriculture 44, 99–110. National Oceanic and Atmospheric Administration National Climatic Data Center, 1997]. Climatological Data Annual Summary, Texas 1997 102:13:71. Peinetti, R., Pereyra, M., Kin, A., Sosa, A., 1993. Effects of cattle ingestion on viability and germination rate of Calden (Prosopis caldenia) seeds. Journal of Range Management 46, 483–486. Rosas, C.A., Engle, D.M., Shaw, J.H., Palmer, M.W., 2008. Seed dispersal by Bison bison in a tallgrass prairie. Journal of Vegetation Science 19, 769–778. SAS, 2002. SAS/STAT Version 9.1 for windows [computer program]. SAS Institute, Cary, NC, USA. Smith, L.L., Ueckert, D.N., 1974. Influence of insects on mesquite seed production. Journal of Range Management 27, 61–65. Steenkamp, H.E., Chown, S.L., 1996. Influence of dense stands of an exotic tree, Prosopis glandulosa Benson, on a savanna dung beetle (Coleoptera: Scarabeinae) assemblage in southern Africa. Biological Conservation 78, 305–311. Tews, J., Schurr, F., Jeltsch, F., 2004. Seed dispersal by cattle may cause shrub encroachment of Grewa flava on southern Kalahari rangelands. Applied Vegetation Science 7, 89–102. Traveset, A., 1998. Effect of seed passage through vertebrate frugivores’ guts on germination: a review. Perspectives in Plant Ecology, Evolution and Systematics 1-2, 151–190. US Department of Agriculture–Natural Resource Conservation Service, 2014]. Web soil survey. Available at: http://websoilsurvey.nrcs.usda.gov/app Accessed 6 March 2014. US Department of Agriculture, Natural Resource Conservation Service, 2014]. Plants database. Available at: http://plants.usda.gov Accessed 6 March 2014.

Valone, T.J., Thornhill, D.J., 2001. Mesquite establishment in arid grasslands: an experimental investigation of the role of kangaroo rats. Journal of Arid Environments 48, 281–288. Van Auken, O.W., 2000. Shrub invasions of North American semiarid grasslands. Annual Review of Ecology and Systematics 31, 197–215. Van Auken, O.W., Bush, J.K., 1997. Growth of Prosopis glandulosa in response to changes in aboveground and belowground interference. Ecology 78, 1222–1229. Van Klinken, R.D., White, A.J., 2011. Overcoming seasonally fluctuating resources: bruchid predation of mesquite (Prosopis) seed in dung. Biological Control 59, 361–365. Villagra, P.E., Marone, L., Cony, M.A., 2002. Mechanisms affecting the fate of Prosopis flexuosa (Fabaceae, Mimosoideae) seeds during early secondary dispersal in the Monte Desert, Argentina. Australian Ecology 27, 416–421. Watts, J.G., Hewitt, G.B., Huddleston, W., Kinzer, H.G., Lavigne, R.J., Ueckert, D.N., 1989. Rangeland Entomology. Society for Range Management Range Science Series, No. 2, Denver, CO, USA. Weltzin, J.F., Archer, S., Heitschmidt, R.K., 1997. Small-mammal regulation of vegetation structure in a temperate savanna. Ecology 78, 751–763. Wolverton, S., Kennedy, J.H., Cornelius, J.D., 2007. A paleozoological perspective on whitetailed deer (Odocoileus virginianus texana) population density and body size in central Texas. Environmental Management 39, 545–552. Wright, H.A., Bunting, S.C., Neuenschwander, L.F., 1976. Effect of fire on mesquite. Journal of Range Management 29, 467–471. Zar, J.H., 1996. Biostatistical analysis. third ed. Prentice Hall, Upper Saddle River, NJ, USA 662 pp. plus Appendices. Zimmermann, H.G., 1991. Biological control of mesquite, Prosopis spp. (Fabaceae), in South Africa. Agriculture, Ecosystems & Environment 37, 175–186.

Please cite this article as: Ansley, RJ., et al., Mesquite Pod Removal by Cattle, Feral Hogs, and Native Herbivores, Rangeland Ecology & Management (2017), http://dx.doi.org/10.1016/j.rama.2017.01.010