Open Access
How to translate text using browser tools
20 September 2022 Long-Term Effects of Cattle Ranching on Soil Nitrogen and Phosphorus Balances in a Savanna Ecosystem
Peter Edwards, Patrick Cech, Judith Sitters, Harry Olde Venterink
Author Affiliations +

We used ranch records and soil analyses to investigate the effects of cattle ranching on phosphorus (P) and nitrogen (N) balances in a humid, tallgrass savanna ecosystem in Tanzania. Over a 41-yr period between 1958 and 1999, the ranch supported an average of 10 435 cattle. These consumed an estimated 571 586 tons (t) of dry matter containing 692 t P and 6 230 t N. Of these nutrients, 162 t of P (23%) and 602 t of N (9.7%) were exported in animals leaving the ranch, while 222 t (32 %) P and 2 364 t of N (38 %) were transferred as excreta to the night corrals. The measured excesses of nutrients in the soil of the corrals were equivalent to 59% of all P and 19% of all N deposited in these areas over 41 yr. Total losses from the pastures amounted to 10.2% of P and 6.6% of N in the top 20 cm of tall grass savanna soil. These losses, especially of P, probably reduced the nutritional quality of the pastures and may have contributed to the reported decline in animal fertility. In addition, they may have promoted the spread of secondary woodland dominated by Vachellia (formerly Acacia) zanzibarica. Three general conclusions can be drawn from this study. First, humid tallgrass savannas on nutrient-poor soils are unsuitable for intensive livestock production. Second, over an extended period the loss of nutrients from cattle pastures can be ecologically significant. Ensuring the sustainability of grazing systems requires measures to counteract this loss, such as the use of shifting night corrals. Third, ranch records, while lacking the precision and detail possible in experimental studies, can provide valuable insights into long-term effects of ranching that would be difficult to obtain by other means.


Effects of large herbivores on nutrient cycling are complex. Animals can create spatial patterns in vegetation through their choice of feeding sites and by influencing the local turnover of nutrients; and if animals feed and drop their excreta in different places, they may also create distinctive areas of nutrient enrichment and depletion (Schnyder et al. 2010; Shibui and Wegener 2010; Vuorio et al. 2014). In a traditional Alpine pasture in Switzerland, for example, high amounts of phosphorus (P) were found to accumulate in small areas of level ground where cattle rested at night, while much larger areas were depleted (Jewell et al. 2007). The authors concluded that this process, operating over a period of centuries, had reduced the productivity of the pasture until it was no longer economically viable.

Local concentrations of nutrients can be even higher if animals are confined overnight in enclosures (variously referred to as corrals, paddocks, kraals or bomas), which is a common practice in African savannas. Several studies in savanna vegetation have demonstrated how large-scale patterns can develop through nutrient transfer by cattle (Stelfox 1986; Augustine 2003; Tobler et al. 2003; Cech et al. 2010; Kioko et al. 2012; Porensky and Veblen 2015; Marshall et al. 2018). In one such study, of a traditional pastoral system at Mpala Ranch in central Kenya, Augustine (2003) found that abandoned bomas remained enriched with nutrients, especially P, for several decades, and became important grazing areas for wildlife. Examples like this have led some authors to suggest that nutrient-enriched patches can be beneficial for the grazing system as a whole; Porensky and Veblen (2015), for example, argue that concentrating livestock in corrals creates ecosystem hotspots that increase the functional heterogeneity of the landscape and attract wildlife, as well as providing palatable forage for livestock. One management recommendation to counteract the depletion of nutrients from the surrounding grazing areas, which has been widely reported in African savanna (Augustine 2003; Cech et al. 2010; Van der Waal et al. 2011; Sitters et al. 2020), is to relocate such corrals at regular intervals.

Most studies of cattle grazing in African savanna have been of traditional pastoral systems, and much less is known about the impacts of large-scale commercial ranching. Also, while many studies have investigated the spatial patterns of soil nutrients resulting from cattle grazing (e.g., in cattle resting sites, corrals, and grazing grounds), none has quantified nutrient flows at a ranch scale, which would be useful for understanding long-term impacts of cattle ranching on ecosystem properties. Such a study should obviously include both the transfer of nutrients within the ranch and the export of nutrients in livestock and animal products (“harvested”). Because such data are mainly lacking for African savanna, it is difficult to evaluate the impact of cattle ranching in these ecosystems or advise on how to manage ranching enterprises sustainably.

We studied the transfer of nutrients by cattle on the former Mkwaja Ranch on the coast of Tanzania (Fig. 1). This ranch, established in 1954, represented an ambitious attempt to make a profitable business out of large-scale livestock production in a nutrient-poor savanna ecosystem. However, the operation was dogged by increasing problems of bush encroachment and declining productivity, leading to its closure in 2000. The ranch kept comprehensive records about all aspects of management, including details of stock numbers, growth rates, and fertility, and partly for this reason it was the subject of many research studies. Some of these concerned livestock productivity, fodder resources, and disease control (Ford and Blaser 1971; Gates et al. 1983; Trail et al. 1985), while others focused on the ecology and biogeochemistry of the savanna ecosystem (Tobler et al. 2003; Treydte et al. 2006; Cochard et al. 2014). The spatial redistribution of nitrogen (N) and phosphorus (P) by cattle was studied by Cech et al. (2010), who measured high nutrient concentrations in and around the night corrals. These authors also estimated losses of N and P due to fires (which are frequent during the dry season), inputs of these nutrients in wet atmospheric deposition, and N inputs through N2-fixation (Fig. 2). Other studies at Mkwaja Ranch have investigated the spread of Vachellia (formerly Acacia) zanzibarica in heavily grazed areas, which became a serious problem because it reduced the area of open grazing. Both Cech et al. (2010) and Sitters et al. (2013, 2015) showed that N2-fixation by V. zanzibarica was a major source of available N in an otherwise N-poor ecosystem. In addition, Sitters et al. (2013, 2015) reported large increases in soil organic P beneath V. zanzibarica, probably because the deep-rooted trees could tap P that was not available to grasses (see Fig. 2).

In this study, we used ranch records to estimate total quantities of nutrients transferred by livestock over the 41-yr period that the ranch was in full operation (1958–1999). Because it had closed by the time of our study, we were unable to use Augustine's (2003) method of measuring the quantities of nutrients in excreta, so we estimated nutrient flows indirectly from livestock production data. To do this, we calculated the total forage intake by all cattle kept on the ranch, using published equations that relate forage intake by beef cattle to liveweight and rate of growth. We then used data on the nutrient content of forage and livestock (i.e., N and P contents of mature cows) to calculate the total uptake of nutrients removed from the pastures and the quantities lost from the ecosystem in both excreta returned to the night corrals and in animals leaving the ranch. Finally, we compared deposition of nutrients in excreta, as estimated from our forage intake calculations, with the measured increment in pools of soil N and P in the corrals and estimated the corresponding decline in these nutrients in the surrounding pastures. In the discussion, we consider how nutrient transfer from grazing areas affected the savanna ecosystem and viability of the ranching operation.

Figure 1.

Map of the research area. Saadani National Park with the former Mkwaja Ranch (indicated in light gray) in the North and the former Saadani Game Reserve (indicated in dark gray) in the South. The former corrals where cattle were held at night are indicated with closed circles. Arrows show the directions of three transects along which soil samples were collected. Filled triangles are villages.



Study area

The former Mkwaja Ranch (headquarters at 5°44″S, 38°48″E) is located in Saadani National Park on the Tanzanian coast, directly opposite the island of Zanzibar, and occupies an area of 462 km2 (see Fig. 1). To the south it shares a border with the former Saadani Game Reserve (hereafter Saadani), which protects some 1 178 km2 (5°43″S, 38°47″E) of undisturbed coastal vegetation and its wildlife. The landscape in this region is composed of open grasslands interspersed with evergreen thickets and in some places coastal forest. Since rainfall is high enough to support forest, the grassland is probably of anthropogenic origin, being the product of fires over many centuries (Cochard et al. 2014). This vegetation is dominated by tall (< 2.5 m), tussock-forming C4-grasses such as Hyperthelia dissoluta, Diheteropogon amplectens, and Andropogon schirensis, together with the palm Hyphaene compressa (Fig. 3a).

Figure 2.

Important fluxes of N and P in the Mkwaja Ranch grazing system (g · m–2 · yr–1). Solid arrows and bold font are fluxes investigated in this study. Hatched arrows are fluxes estimated by Cech et al. (2010); losses through fire are weighted mean values assuming that acacia woodland comprises 38% of the total grazing area. Unfilled arrows are fluxes for which no data are available.


Rainfall varies greatly from year to year, with annual precipitation measured at Mkwaja Ranch between 1957 and 1998 ranging from 610 to 1 700 mm, with a mean of 1 040 mm. The wet season lasts from March until June, and there is a short rainy season from mid-October to mid-November. The driest months are January to February and August to September. Fires, many started deliberately, are common during dry periods.

The soils in the study area are mostly gray vertisols derived from coral sands. They consist of mainly grayish fine sand or loamy sand, except in low-lying areas where black clay soils predominate, and are relatively unstructured to a depth of > 1.5 m. Concentrations of soil nutrients, notably P, are generally much lower than in most semiarid savannas (Cech et al. 2008), and the nutritional quality of grasses for ungulates is mainly poor (Stähli et al. 2015). In a study of resource partitioning by native ungulates in Saadani, Stähli et al. (2015) found that animals scarcely used most grass species, and that only two, Panicum infestum and Digitaria milanjiana, consistently met the minimum N and P requirements of small- and medium-sized grazers. The authors concluded that the low quality of herbage in these wet, oligotrophic savannas explained the impoverished grazing community at Saadani (and also Mkwaja), which is dominated by just two native ungulates, Bohor reedbuck Redunca redunca and waterbuck Kobus ellipsiprymnus.

In 1953, Amboni Estates Limited acquired the land that was to become Mkwaja Ranch and began to stock it with East African Zebu (Bos taurus indicus) cattle and Boran bulls (Ford and Blaser 1971). The Boran breed, which is predominantly Zebu but with some taurine background, is widely used in Tanzania and Kenya, where it is valued as a hardy, low-maintenance breed that can feed efficiently on low-quality forage. Breeding at Mkwaja Ranch was managed using both ranch-bred bulls and semen of Boran bulls from the Kabete Artificial Insemination Centre, Kenya, and in the mid-1980s the herd was considered to be “grade Boran” (Trail et al. 1985).

Over the years, the ranch developed into one of the largest private cattle enterprises in Tanzania, supporting up to 15 000 head of livestock during the 1970s (see Fig. 3a). After an initial period when various management schemes were tried, the grazing land was organized in 1958 into sections, of which there were up to 23. Each section supported several herds of 150 to 200 animals and was centered on a large corral (> 4 ha) where up to 1 500 cattle were kept at night (see Fig. 3b). At around 6 A.m. each morning, the various herds were led from the corrals to pasture areas and later to a nearby dam. Initially, the choice of grazing areas was relatively haphazard, but in 1976 a rotational system was introduced to prevent the most accessible areas from being overused (Ford and Blaser 1971; Gates et al. 1983; Trail et al. 1985). Even so, ranch records reveal that some parts of the ranch were used more intensively than others; for example, half of the corrals, mainly in the northern sector of Mkwaja Ranch, accounted for 80% of all corral usage. Steers and any heifers not retained as breeding cows were sold at an age of about 45 mo. In addition, many heifers and steers were taken from the ranch to neighboring sisal estates for fattening (Trail et al. 1985).

Trypanosomiasis transmitted by tsetse flies (Glossina spp.) was a major problem, to the extent that cattle could not survive at Mkwaja Ranch unless treated with trypanocidal drugs—initially with the prophylactic drug Antrycide prosalt (Imperial Chemical Industries Ltd.) and from 1965 onwards with Samorin (isonetamidium chloride; May and Baker Ltd.). In a major study conducted in the early 1980s, Trail et al. (1985) found that the productivity of cattle at Mkwaja treated with samorin was approximately 80% that of Boran cattle reared in a tsetse-free ranching environment in Kenya, but 35% higher than that of trypanosome-tolerant N'Dama cattle kept without trypanocidal drugs. The study by Trail et al. (1985) provided a wealth of information about the management and productivity of the Mkwaja herd over a 10-yr period, including calving rate, calf weight at weaning, mean weight of mature cows, and growth rates (Table 1). These data enabled us to calculate the total dry matter intake on the ranch.

Figure 3.

Typical savanna vegetation in the former Mkwaja ranch (a) and a corral with Boran cattle (b).


The effects of ranching on large-scale vegetation patterns at Mkwaja Ranch were investigated by Tobler et al. (2003), who used a combination of field survey and remote sensing to describe the vegetation and compare it with an adjacent area (Saadani) that had never been used for ranching. This study demonstrated more or less concentric zones of distinct vegetation at different distances around the former corrals (Tobler et al. 2003). These zones, described in more detail in Cech et al. (2010), included 1) a narrow zone of highly disturbed vegetation extending to about 100 m outside the corrals (referred to here as the corral margins); this vegetation was dominated by the grasses Digitaria milanjiana, Eragrostis superba, and the sedge Cyperus bulbipes; 2) Heteropogon savanna, dominated by the grass Heteropogon contortus and accompanied by several other species; 3) secondary woodland (referred to here as acacia woodland), dominated by Vachellia zanzibarica, which extended between 300 and 2 500 m from the corral, with the highest densities occurring at 900 m; and 4) tallgrass savanna dominated by the grasses Hyperthelia dissoluta and Diheteropogon amplectens (Cech et al. 2010); this was the dominant vegetation beyond 2 500 m from the corrals. Of these zones, tallgrass savanna occupied the largest area and provided the main feeding grounds for livestock.

Despite great efforts to adapt livestock production to local conditions, the ranch experienced continuing problems of disease, bush encroachment, and declining fertility and was never profitable (Gross et al. 2005). It closed in 2000.

Nutrient intake and redistribution

From the annual ranch reports from 1958 until 2000 we extracted data on numbers of cows with calves, cows without calves, fattening animals, and animals sold (see Table 1;  supplementary materials (mmc1.xlsx) ). We omitted data for the first 4 yr of ranching (1954–1957), when numbers of cattle were small and there was no corral system.

To calculate dry matter intake (DMI, kg · d–1) by cattle, we used an equation developed for Nellore (Zebu) beef cattle on pasture (Valadares Filho et al. 2016, equation 2.6):


where SI is supplementary intake (none at Mkwaja Ranch), BW is body weight (kg), and ADG is average daily gain (kg · d–1). For heifers and steers, we used the ADG values in Trail et al. (1985) to calculate mean DMI values from weaning at 8 mo until 45 mo, when steers and any surplus heifers were sold. For breeding cows, we estimated mean DMI over the production cycle (i.e., the period from one calving to the next) by multiplying the DMI requirement for maintenance (4.63 kg for a 286-kg cow) by a factor representing the additional energy needs of pregnancy and lactation (EPL). To calculate EPL, the following terms were required (Valadares Filho et al. 2016, Table 10.6; details in  supplementary materials (mmc1.xlsx) ):

  1. Metabolizable energy (ME; Mcal · d–1) for maintenance (MEm) over a production cycle of 567 d, calculated as MEm = (130 × EBW0.75)/1 000 (Valadares Filho et al. 2016, equation 10.7), where EBW is empty body weight (assumed to be 0.864 × BW).

  2. Additional ME for pregnancy (MEp) over 285 d, calculated as MEp = CBW × 0.000000793 × TG3.017/(1000 × 0.12), where CBW is mean calf weight at birth and TG is days pregnant (Valadares Filho et al. 2016, equation 10.14). To simplify the calculation, pregnancy was divided into early, mid, and late phases, and average values were calculated for the mid and late phases (mid-phase: 95 d centered on d 191; late phase: 60 d centered on d 264). MEp in the early phase was small and ignored.

  3. Additional ME for lactation (MEl) over 240 d, calculated as 0.499 × MEm, derived from data for Zebu cows in Calegare et al. (2009).

Table 1

Parameters used to calculate dry matter intake and nitrogen and phosphorus transfer by cattle at Mkwaja Ranch.


The factor EPL was then calculated using the expression EPL = (567 × MEm + 95 × MEp(d191) + 60 × MEp(d264) + 240 × MEl)/(567 × MEm). We obtained a value of 1.28 for EPL, which yielded a mean DMI of 5.92 kg for breeding cows on Mkwaja Ranch.

To estimate nutrient intake in dry matter, we used average concentrations of nutrients in regrowth of tallgrass savanna vegetation from Cech et al. (2008).

To estimate the N and P contents of animals leaving the ranch, we used concentration data for Nellore beef cattle similar in size to the Boran cattle at Mkwaja (Valadares Filho et al. 2010). Because that publication gave separate concentrations for empty body weight and for organs and viscera, we calculated weighted mean values for the entire animal using data on proportions also given in Valadares Filho et al. (2010) (see Table 1; details in  supplementary materials (mmc1.xlsx) ). Many animals were sent to neighboring sisal estates for fattening, and we therefore applied a factor to account for the mean proportion of postweaning body weight gained on the ranch. This factor was based on estimated numbers of fattening animals (assuming that the animals were sold at 45 mo) and the number actually present on the ranch (ranch records).

The remaining nutrient (i.e., difference between intake and export) was assumed to be returned in excreta. We assumed that 42% of nutrients were returned between 18.00 and 6.00, while the cattle were in the night corrals. We based this value on four published studies showing diurnal trends in defecation frequency in cows (see Table 1;  supplementary materials (mmc1.xlsx) ).

N and P pools in the soil

Soil samples for chemical analysis were collected in 2003, some 3–4 yr after the last cattle had left the ranch. For three former corral systems (Kichangani, Mariamu, and No. 6; see Fig. 1), we selected five locations representative of the corral center, the corral margin (taken as a zone extending ≈100 m from the former corral fence) and tallgrass savanna (Cech et al. 2010). At each site, we used a soil corer to collect five 70-cm cores and divided them into five segments (0–10, 10–20, 20–30, 30–50, 50–70 cm). For each depth, the five segments were pooled. Soil bulk density was determined by digging a pit and extracting two large soil cores (4.2-cm diameter) per depth segment that were then dried to constant weight. Total C and N in soil samples were measured using a dry combustion analyzer (CN-2000, LECO Corp., St. Joseph, MN). Total soil P was measured after Kjeldahl digestion using a continuous-flow injection analyzer (FIAStar, FossTecator, Höganäs, Sweden), as described in Cech et al. (2010).

Table 2

Estimated quantities of N and P ingested by cattle on Mkwaja Ranch between 1958 and 2000 and the amounts returned in excreta and exported in animals leaving the ranch.


To extrapolate our findings to the entire ranch (i.e., 23 corral systems), we used an estimate of the total area of corrals derived from a multispectral classification of Landsat TM images (Tobler et al. 2003). The corral areas can readily be detected in these images, and ground-truthing studies indicated a high level of accuracy in classifying this particular habitat (totaling 92 ha). The area of the corral margin was taken to be a 100-m fringe surrounding each corral (totaling 276 ha). In calculating the total quantities of N and P in all corrals, we applied a weighting factor to account for the higher than average use of the three corrals we studied (relative corral use/relative grazing area = 19%/13% = 1.46).

In estimating the excess N and P in corrals and corral margins, we calculated soil nutrient pools for the top 70 cm of soil (Cech et al. 2010), to be sure to include any nutrients that had leached into deeper soil. In addition, we calculated soil N and P pools for the top 20 cm, to represent the nutrients in the main rooting zone of grasses.


In the 41 yr between 1958 and 1999, the average number of cattle (including suckling calves) on Mkwaja Ranch was 10 435 (Fig. 4a). These animals consumed an estimated 571 586 tons of dry matter containing 692 t P and 6 230 t N (Table 2). Of these amounts, 23% of P (162 t) and 9.7% of N (602 t) were exported in animals that left the ranch, while the remainders (529 t P and 5 629 t N) were assumed to be returned in excreta. The estimated quantities transferred to the corrals were 222 t P and 2 364 t N. Calculated per unit area of grazing land, the average intakes of nutrients were 0.0487 g P · m–2 · yr–1 and 0.439 g N · m–2 · yr–1. The total quantities lost from the grazing areas, either in animals that left the ranch or in excreta deposited in corrals, amounted to 0.027 g P · m–2 · yr–1 and 0.208 g N · m–2 · yr–1.

Nutrient pools in the top 70 cm of tallgrass savanna soil were 36.5 g P · m–2 and 358 g N · m–2 in tallgrass savanna and 243 g P · m–2 and 1 075 g N · m–2 in the highly enriched soil of corral areas (Table 3). P concentrations in the corral centers declined rapidly below a depth of about 50 cm to levels similar to those in the tallgrass savanna, suggesting that 70 cm was an adequate depth to capture most of the additional P introduced by cattle. Indeed, extrapolating the concentrations at 70 cm down to a depth of 2 m only increased the estimated P excess in the corrals by a further 6%. N and P concentrations were far lower in the corral margins than in the corrals and not significantly higher than in tallgrass savanna; the estimated pools were 45.8 g P · m–2 and 493 g N · m–2 (see Table 3).

Figure 4.

Temporal trends in cattle stock and calving rate. a, Cattle stock and b, calving rate in the Mwkaja ranch according to annual ranch reports. Calving rate is the percentage of annually produced calves by the population of cows. The stock increase in 1991 was due to cattle imported from elsewhere in a final attempt to make the ranch profitable.


Table 3

Mean (± standard error) pools of total phophorus and nitrogen in soil of corrals, corral margins, and tallgrass savanna.


Extrapolating these values to the 23 corrals on the ranch, the total excesses in corral were calculated to be 130 t P and 452 t N, or 59% of all P and 19% of all N transferred to corrals in excreta over 41 yr (see Table 2). Including the corral margins, which were also somewhat enriched, the total excesses were 148 t P and 707 t N, or 66% of P and 30% of N transferred over 41 yr.


The detailed ranch records kept by Mkwaja Ranch, combined with our soil analyses, provide a unique opportunity to assess the long-term consequences of cattle grazing on a nutrient-poor savanna ecosystem. Our “back-of-the-envelope” calculations clearly demonstrate that ranching at Mkwaja was not sustainable. We estimate that nearly one third of all P consumed by cattle was exported from the ranch and a further 29% was transferred to the corrals, where trampling by cattle prevented any vegetation from growing. Over 4 decades, this export and immobilization of nutrients led to substantial reductions in soil N and P pools and availability, as was also demonstrated for a long-term exclosure experiment at Mpala Ranch in Kenya (Sitters et al. 2020). These results suggest that this type of cattle ranching—with no spatial redistribution of the corral nutrients and with a severely reduced wild megaherbivore population—leads to soil impoverishment in the grazing areas and eventually a collapse in cattle production.

Before discussing our findings in more detail, it is necessary to consider the accuracy of the nutrient-flux estimates. The stock and productivity data on which our estimates were based are probably accurate, but we had to make many assumptions in applying them to the entire ranch. One potential source of error was the mean nutrient content of forage, which varied widely among vegetation types and seasons (Kozak 1983; Cech 2008). For example, the P concentration of grass regrowth in acacia woodlands was 0.083% (dry weight), while it was 0.121% in tallgrass savanna and 0.189% in corral margins. Similarly, N concentrations ranged from 1.02% in the corral margin to 1.40% in acacia woodland. We used the values for tallgrass savanna because it was the most extensive vegetation type and probably most important in the diet of cattle. Nutrient concentrations for this vegetation compare well to similarly low levels for forage in other tropical areas (McDowell 1985). A second potential source of error was the equation used to estimate DMI, which was based on experimental studies using Nellore cattle in Brazil. This error is probably small (< 10) because both Nellore and Boran are breeds of Zebu cattle adapted to nutrient-poor pasture and the Nellore animals were only a little larger than the Boran at Mkwaja Ranch. More significant, perhaps, is the implicit assumption that animals can feed ad libitum, whereas at Mkwaja there was sometimes a serious shortage of forage in the dry season. To investigate how this could have affected the result, we modeled DMI of fattening animals assuming that average daily gain was zero for a 3-mo period each yr. With this seasonal model, 4% more DMI was needed to reach the same final body weight, suggesting that our estimate may be a little too low. A third possible source of error is in the proportion of nutrients returned in excreta to the night corrals, which we took to be 42%. We consider that any error arising from this assumption was probably small (< 10%) because the proportion of dung dropped at night varies rather little among studies (38–44%;  supplementary materials (mmc1.xlsx) ). On the other hand, the proportion of urine excreted at night, which we assumed to be the same as for dung, is more variable (32–49%). However, this would only have affected our N estimates, since cows feeding on nutrient-poor forage excrete little P in their urine (Sitters and Olde Venterink 2021).

Despite these uncertainties, we consider that our estimates of nutrient flow are probably accurate to within 20% at worst. Support for this assumption is the similarity between our estimate of P redistributed in excreta and the excess soil P present in corrals and corral margins (222 vs. 148 t P). This result suggests that 45 yr after the corral system was introduced, in 2003, two thirds of the P transferred to corrals and corral margins remained in the upper soil layers, as has been found in other studies (Augustine 2003). The fact that much less N was retained in the soil is unsurprising because N is readily lost from excreta through volatilization, especially from urine, and from the soil through nitrate leaching (Augustine 2003).

Mkwaja Ranch differed in important respects from most other savanna ecosystems where nutrient transfer has been investigated. Most of these were relatively nutrient-rich, semiarid systems that would formerly have supported a high diversity and biomass of wildlife. Mkwaja, in contrast, was a humid, fire-maintained grassland, almost certainly of anthropogenic origin, and was probably depleted of nutrients even before it became a ranch (Cochard et al. 2014). Its impoverished state becomes clear by comparing the nutrient pools in tallgrass savanna with those in the remaining forest fragments at Mkwaja (10.9 vs. 41.8 g P · m–2 in the top 20 cm of soil; Table 3, Cech et al. 2010) or with those in Augustine's (2003) site at Mpala Ranch in Kenya (8.2 vs. 20 g P · m–2 in the top 15 cm of soil). Recent data from Mpala show that 20 yr of cattle grazing had reduced soil P (10 g P · m–2 in comparison with 12–16 g P · m–2 in sites without cattle) and N pools, which was reflected in lower grass nutrient concentrations (especially of N; Sitters et al. 2020). Indeed, Stähli et al. (2015) concluded that poor nutritional quality of forage was the reason for the low densities and restricted diversity of native grazers in the neighboring Saadani National Park (cf. Fig. 1).

Despite these inauspicious circumstances, the ranch owners at Mkwaja made determined efforts to use the land for commercial beef production, keeping livestock at far higher densities than is usual in traditional pastoral systems (up to 29.5 cattle km–2, compared with 7.3–16.3 cattle km–2 at Mpala Ranch; Augustine 2003). Furthermore, they set up permanent corrals to hold animals at night, so the ranch gained none of the benefits of shifting enclosures. We estimate that the mean annual losses of N and P from grazing areas were 0.027 g P · m–2 and 0.208 g N · m–2, respectively, of which 0.0156 g P · m–2 and 0.166 g N · m–2 were deposited in excreta in the corrals (see Fig. 2; Table 3). In the case of P, the transfer to corrals was somewhat lower than estimated for Mpala Ranch (0.021–0.026 g P · m–2 · yr–1; Augustine 2003); because the Mkwaja soils were nutrient poor and the corrals permanent, however, the impact on the ecosystem may have been greater.

The accumulated losses from grazing areas over 41 yr were equivalent to 10.2% of soil P and 6.6% of soil N (top 20 cm). However, these are average values for the entire ranch; since the northern half of the ranch was used more intensively than the southern half (and as a consequence also suffered more from woody encroachment, Tobler et al. 2003), the losses from grazing land in this area were probably at least twice as high as those for the ranch as a whole. The P removal was probably most significant for the ecosystem because less than half of total soil P would have been in an organic form potentially available to plants (Sitters et al. 2013). However, because leguminous plants were scarce and estimated levels of N2-fixation low (see Fig. 2; Cech et al. 2010), N depletion may also have reduced grassland productivity. On the other hand, these losses may have been mitigated by nutrients entering the ecosystem in atmospheric deposition. Cech et al. (2010) reported a P input of 0.033 g · m–2 · yr–1, which was similar to the average loss through grazing, and an N input of 0.48 g · m–2 · yr–1, which was about twice the loss due to grazing. Another factor could have been the reduced frequency and intensity of fires during the ranching period, when there was less biomass to burn and any fires were controlled. Just how important these two processes were in compensating for the losses due to grazing is difficult to assess, however, because neither proportion of atmospheric deposition retained by the ecosystem nor the frequency of fires during the ranching period are known.

Despite these uncertainties, two lines of evidence suggest that the loss of nutrients from grazing areas did have an impact on the savanna ecosystem. One was the marked decline in breeding rate on Mkwaja Ranch, from > 80% in the early 1960s to < 50% in 1985 (see Fig. 4b). It is unlikely that this decline was caused by disease, which could be controlled at a level considered “acceptable” using drugs (Trail et al. 1985). In the period 1973 to 1982, for example, annual mortality of cows due to disease was 4.2%, with anaplasmosis being the most important cause of death (Trail et al. 1985). On the other hand, phosphorus deficiency was recognized as a problem from as early as the 1960s. Commenting in the annual ranch report for 1972/1973, the manager wrote: “There is only one reason [for the low fertility] and that is the low nutritional level in our breeding stock which is exposed to severe seasonal nutritional stresses mainly after calving, while maintaining a healthy calf and at the same time experiencing a fast deterioration of the nutritional value of the pastures due to dying out. … There is a rapid loss of weight and condition of the dam, which is responsible for an extension of the re-breeding period, in fact for missing entirely the next conception, which falls exactly into a time when nutritional stress is highest. The overall result is low productivity not only in cows but in heifers and fattening stock as well” (Gross et al. 2005).

The second line of evidence was the encroachment of grazing areas by secondary woodland composed of Vachellia zanzibarica. This became a significant problem within a few years of opening the ranch and remained so until the ranch closed. The reasons for its spread were complex (Cochard et al. 2014), but two important factors may have been its capacity to fix atmospheric N2 and to obtain P not available to grasses (see Fig. 2). The latter was demonstrated by Sitters et al. (2013), who found that organic P pools increased strongly beneath V. zanzibarica, which they attributed to the trees being able to access P from deeper soil layers and perhaps also to use organic forms of P more efficiently. These two factors would have given V. zanzibarica an increasing competitive advantage over grasses as the soils were depleted of P and N through grazing.

Management Implications

Our results suggest that humid tallgrass savannas on nutrient-poor soils are not suitable for intensive livestock production. Strong contraindications for this form of land use at Mkwaja Ranch include the low nutritional quality of herbage, small pools of N and P in the soil, and low abundance and diversity of native grazers (Stähli et al. 2015). However, if this type of savanna is to be used for ranching, it is clearly important to manage the system so as to minimize the negative effects of nutrient transfer. In this respect, we echo a conclusion reached in several previous studies—that continuous cattle grazing of rangelands has negative consequences for the ecosystem and forage production, whereas more flexible grazing methods, for example with shifting bomas or corrals, can even have beneficial effects (Stelfox 1986; Kioko et al. 2012; Porensky and Veblen 2015; Sitters et al. 2020).

Another interesting possibility might be to accept encroachment by acacia woodland as an inevitable consequence of grazing nutrient-poor savanna and take advantage of it to restore the depleted soils. At Mkwaja Ranch, the spread of woodland was a serious management problem because it reduced the area available for grazing and was costly to control. However, the trees are relatively short-lived and do not regenerate in the absence of grazing. It might be possible to develop a long-term rotational system in which only part of the ranch is used for grazing, while the remainder is set aside as a woodland fallow to restore soil fertility and produce firewood (Cochard et al. 2014). Clearly, research would be needed to determine the commercial viability of such a system, also studying long-term effects of P exploration and depletion from deeper soil layers, but from an ecological perspective it would appear feasible.

Declaration of Competing Interest


Supplementary materials

Supplementary material associated with this article can be found, in the online version, at  doi:10.1016/j.rama.2022.05.006.


We acknowledge Benjamin Donald and John Williams for assistance in the field and Kristel Perreijn for help with Figure 1.



Aland, A., Lidfors, L., Ekesbo, I., 2002. Diurnal distribution of dairy cow defecation and urination. Applied Animal Behavioral Science 78, 43–54. Google Scholar


Auerswald, K., Mayer, F., Schnyder, H., 2010. Coupling of spatial and temporal pattern of cattle excreta patches on a low intensity pasture. Nutritional Cycle Agroecosystems 88, 275–288. Google Scholar


Augustine, D.J., 2003. Long-term, livestock-mediated redistribution of nitrogen and phosphorus in an East African savanna. Journal of Applied Ecology 40, 137–149. Google Scholar


Calegare, L., Alencar, M.M., Packer, I.U., Lanna, D.P.D., 2009. Cow/calf preweaning efficiency of Nellore and Bos taurus × Bos indicus crosses. Journal of Animal Science 87, 740–747. Google Scholar


Cech, P.G., Kuster, T., Edwards, P.J., Olde Venterink, H., 2008. Effects of herbivory, fire and N2-fixation on nutrient limitation in a humid African savanna. Ecosystem 11, 991–1004. Google Scholar


Cech, P.G., Olde Venterink, H., Edwards, P.J, 2010. N and P cycling in Tanzanian humid savanna: Influence of herbivores, fire, and N2-fixation. Ecosystem 13, 1079–1096. Google Scholar


Cochard, R., Edwards, P.J., Weber, E., 2014. Post-ranching tree–grass interactions in secondary Acacia zanzibarica woodlands in coastal Tanzania—an experimental study. Applied Vegetation Science 18, 297–310. Google Scholar


Ford, D., Blaser, E., 1971. Some aspects of cattle raising under prophylactic treatment against trypanosomiasis on the Mkwaja Ranch, Tanzania. Acta Tropica 28, 69–79. Google Scholar


Gates, D.B., Cobb, P.E., Williamson, D.L., Bakuli, B., Dame, D.A., Blaser, E., 1983. Integration of insect sterility and insecticides for control of Glossina morsitans morsitans Westwood (Diptera, Glossinidae) in Tanzania. 3. Test site characteristics and the natural distribution of tsetse flies. Bulletin of Entomology Research 73, 373–381. Google Scholar


Gross, M., Hoffmann-Riem, H., Krohn, W., 2005. Ch. 4. Rekursives lernen in der sackgasse: viehzucht in der savanne Tansanias. Realexperimente-ökologische gestaltungsprozesse in der wissensgesellschaft. Transcript Verlag, Bielefeld, Germany, pp. 79–110. Google Scholar


Hirata, M., Higashiyama, M., Hasegawa, N., 2011. Diurnal pattern of excretion in grazing cattle. Livestock Science 142, 23–32. Google Scholar


Jewell, P.L., Käuferle, D., Güsewell, S., Berry, N.R., Kreuzer, M., Edwards, P.J., 2007. Redistribution of phosphorus by mountain pasture in cattle on a traditional the Alps. Agriculture Ecosystems and the Environment 122, 377–386. Google Scholar


Kioko, J., Kiringe, J.W., Seno, S.O., 2012. Impacts of livestock grazing on a savanna grassland in Kenya. Journal of Arid Land 4, 29–35. Google Scholar


Kozák, A. 1983. The nutritional value of a tropical natural pasture in Tanzania [thesis]. Zurich, Switzerland: ETH Zurich. Google Scholar


Marshall, F., Reid, R.E.B., Goldstein, S., Storozum, M., Wreschnig, A., Hu, L., Kiura, P., Shahack-Gross, R., Ambrose, S.H., 2018. Ancient herders enriched and restructured African grasslands. Nature 561, 387–390. Google Scholar


McDowell, L.R., 1985. Contribution of tropical forages and soil toward meeting mineral requirements of grazing ruminants. In: Cunha, T.J., McDowell, L.R. (Eds.), Nutrition of grazing ruminants in warm climates. Elsevier, Amsterdam, The Netherlands, pp. 165–188. Google Scholar


Porensky, L.M., Veblen, K.E., 2015. Generation of ecosystem hotspots using short--term cattle corrals in an African savanna. Rangeland Ecology & Management 68, 131–141. Google Scholar


Sahara, D., Ichikawa, T., Aihara, Y., Kawanishi, H., Nagashima, M., 1990. Eliminative and reposing behaviour of dairy cows in the stanchion stall barn. Japanese Journal of Zootechnology Science 61, 249–254. Google Scholar


Schnyder, H., Locher, F., Auerswald, K, 2010. Nutrient redistribution by grazing cattle drives patterns of topsoil N and P stocks in a low-input pasture ecosystem. Nutritional Cycles in Agroecosystems 88, 183–195. Google Scholar


Shibui, N., Wegener, H.R., 2010. The significance of cattle for the nutrient patterns. In: Plachter, H., Hampicke, U. (Eds.), Large-scale livestock grazing—a management tool for nature conservation. Springer, New York, NY, USA, pp. 256–270. Google Scholar


Sitters, J., Edwards, P.J., Olde Venterink, H., 2013. Increases of soil C, N, and P pools along an Acacia tree density gradient and their effects on trees and grasses. Ecosystems 16, 347–357. Google Scholar


Sitters, J., Edwards, P.J., Suter, W., Olde Venterink, H., 2015. Acacia tree density strongly affects N and P fluxes in savanna. Biogeochemistry 123, 285–297. Google Scholar


Sitters, J., Kimuyu, D.M., Young, T.P., Claeys, P., Olde Venterink, H., 2020. Negative effects of cattle on soil carbon and nutrient pools reversed by megaherbivores. Nature Sustainability 3, 360–366. Google Scholar


Sitters, J., Olde Venterink, H., 2021. Stoichiometric impact of herbivore dung versus urine on soils and plants. Plant and Soil 462, 59–65. Google Scholar


Stähli, A., Edwards, P.J., Olde Venterink, H., Suter, W., 2015. Convergent grazing responses of different-sized ungulates to low forage quality in a wet savanna. Australian Ecology 40, 745–757. Google Scholar


Stelfox, J.B., 1986. Effects of livestock enclosures (bomas) on the vegetation of the Athi Plains, Kenya. African Journal of Ecology 24, 41–45. Google Scholar


Tobler, M.W., Cochard, R., Edwards, P.J., 2003. The impact of cattle ranching on large-scale vegetation patterns in a coastal savanna in Tanzania. Journal of Applied Ecology 40, 430–444. Google Scholar


Trail, J.C.M., Sones, K., Jibbo, J.M.C., Durkin, J., Light, D.E., Murray, M., 1985. Productivity of Boran cattle maintained by chemoprophylaxis under trypanosomiasis risk. Report No, 9. International Livestock Center for Africa, Addis Ababa, Ethiopia, p. 76. Google Scholar


Treydte, A.C., Bernasconi, S.M., Kreuzer, M., Edwards, P.J., 2006. Diet of the common warthog (Phacochoerus africanus) on former cattle grounds in a Tanzanian savanna. Journal of Mammals 87, 889–898. Google Scholar


Valadares Filho, S. C., Marcondes, M. I., Chizzotti, M. L., Paulino, P. V. R. [eds.], 2010. Nutrient requirements of Zebu beef cattle Br-Corte. 2nd ed. Universidade Federal de Viçosa, Brazil: Suprema Gráfica LTDA, p. 193. Google Scholar


Valadares Filho, S. C., Costa e Silva, L. C., Gionbelli, M. P., Rotta, P. P., Marcondes, M. I., Chizzotti, M. L., Prados, L. F. [eds.], 2016. Nutrient requirements of Zebu and crossbred cattle Br-Corte. 3rd ed. Universidade Federal de Viçosa, Brazil: Suprema Gráfica LTDA, p. 314. Google Scholar


Van der Waal, C., Kool, A., Meijer, S.S., Kohi, E., Heitkonig, I.M.A., de Boer, W.F., van Langevelde, F., Grant, R.C., Peel, M.J.S., Slotow, R., de Knegt, H.J., Prins, H.H.T., de Kroon, H., 2011. Large herbivores may alter vegetation structure of semi-arid savannas through soil nutrient mediation. Oecologia 165, 1095–1107. Google Scholar


Vuorio, V., Muchiru, A., Reid, R.S., Ogutu, J.O., 2014. How pastoralism changes savanna vegetation: impact of old pastoral settlements on plant diversity and abundance in south-western Kenya. Biodiversity Conservation 23, 3219–3240. Google Scholar
© 2022 The Authors. Published by Elsevier Inc. on behalf of The Society for Range Management.
Peter Edwards, Patrick Cech, Judith Sitters, and Harry Olde Venterink "Long-Term Effects of Cattle Ranching on Soil Nitrogen and Phosphorus Balances in a Savanna Ecosystem," Rangeland Ecology and Management 84(1), 54-62, (20 September 2022).
Received: 8 November 2021; Accepted: 19 May 2022; Published: 20 September 2022
African savanna
cattle grazing
livestock production
Back to Top