The study was conducted in the Lake Victoria Basin (Figure 1) in three countries of Rwanda, Tanzania and Uganda which have a high number of livestock in the Basin. Four districts in Uganda (Isingiro, Rakai, Lyantonde and Lwengo), two in Tanzania (Ilemela and Magu) and two in Rwanda (Nyagatare, Gatsibo, Rwamagana, Kayonza, Ngoma and Kirehe) were included in the study. Geographically, the study districts are located along the Equator between longitudes 30˚00'E and 34˚00'E, and latitudes 0˚00'S and 4˚00'S.

Following their altitudinal differences and their closeness to the equator, the study areas experience varied climatic conditions. For example, Rwanda has a tropical highland climate, with a temperature profile strongly dependent on elevation and receives bimodal rainfall distribution (from mid-September to mid-December and from March to May) dominated by the progression of the Inter-Tropical Convergence Zone (ITCZ). Nevertheless, the study districts in Rwanda fall under semi-arid climate conditions (owing to its position in the rainy shadow of the western highlands―typical for central and eastern part of the country) with total annual rainfall averaging to about 1000 mm and annual mean daily temperature of about 22˚C. The study districts in Tanzania receive rainfall averaging to about 800 mm per annum with the annual mean temperature ranging between 24˚C and 30˚C. In Uganda, the areas receive on average rainfall ranging between 750 and 1200 mm annually. Rakai District which is located just at the Equator receives more rains as compared to the rest (Table 1).

Land use and cover maps in the study areas of Uganda, Rwanda and Tanzania show a conspicuous reduction in grazing areas and increase in cultivated areas over the last twenty years (Figure 2; Figures 3(a)-(c)). Grazing areas reduced by 18.5%, 25.5% and 30.4% in Uganda, Rwanda and Tanzania between 1986 and 2013, 1990 and 2010 and 1984 and 2010 respectively. In the same period, cultivated areas increased by 11.5%, 24.9% and 12.2% of the total study area in Uganda, Rwanda and Tanzania respectively. The overall reduction in natural vegetation cover (grasslands, bushland, woodland, forests and riverine vegetation) and increase in cultivated areas signifies that cultivation is the major land use type decimating grazing areas. In Tanzania, 65.3% of riverine vegetation, the major grazing area, was converted into cultivation and settlement between 1986 and 2000; and 46.9% converted between 2000 and 2010 to the same types of use [10] . [8] also reported similar trends in Uganda with 5.8%

of grassland converted to cultivation between 1986 and 2000 and 34.8% grasslands converted to cultivation between 2000 and 2013. In Rwanda, 15% of grazing areas were converted into other use types between 1990 and 2000 while 12.4% were converted between 2000 and 2010. Noticeable increase in forest areas were also revealed in later period than the former for both study areas in Rwanda and Uganda. The increase in forest in Rwanda for the period 2000-2010 has been due to the regeneration caused by stopping encroachment in the Akagera National Parks that constitutes a larger part of the study area. The stopping encroachment was an outcome of end of emergency period in Rwanda, where the encroachment in the newly redefined boundaries of the Akagera National parks started to be controlled with electrical fences and ranger patrol. Available information revealed that over 50% of the Akagera National Park was encroached during the emergency period.

The tremendous decline in grazing areas may be greatly attributed to high increase in human population within the basin (Figure 4(a), Figure 4(b); Table 2). Between 1986 and 2013, the human population in five districts of Uganda under this study increased by 61.2% from 963,456 to 1,552,900. The increasing population creates more pressure on land for settlement as well as cultivation to produce more food. Similar trends in population growth around the Lake Victoria Basin have been reported by [1] that the LVB is the most densely populated and with the highest growth rates in the region.

The decline of wetland, bushland and grasslands areas has caused the decrease in pasture as the extensive production system, which is practiced by beef cattle keepers in the study area, is heavily dependent on natural grasslands as the source of feed for cattle. Following increase in human population (highest in Africa) in Rwanda, most of the pastureland areas have been converted to agriculture thus shrinking the available pastureland for beef cattle production. With the extreme population density, the mid and high-altitude, farmers had

started exploring the possibility of immigrating to the lower eastern savanna and thus introducing more crop based farm management in this area. The government induced some changes in the land use by promoting rice production in the swamps which attracted more crop based farmers, while animal movement was controlled with zero grazing in mid and high altitude, such policy started to be enforced around 2005. The reduction in grazing area has resulted to inadequate level of nutrition which is the major limiting factor to beef cattle production, poor performance in reproduction and weight gain.

The climate of the study areas in the Lake Victoria Basin was found to vary with time. Table 3 and Table 4 present trends of seasonal and annual rainfall and temperature over time in Tanzania and Uganda study sites. A similar analysis was not possible for Rwanda due to lack of long time series of monthly rainfall and temperature, nevertheless the average historical monthly rainfall and temperature for different periods obtained from online climate databases are presented.

The results on trend analysis for both rainfall and temperature revealed variability between the periods under investigation (Table 3 and Table 4). The slopes of the trend lines for short rains; September, October and November (SON) and long rains; March, April and May (MAM) at the study stations in Tanzania and Uganda revealed a declining trend with the exception of Mwanza airport station. The analysis using linear regression and 5 years moving averages in Uganda revealed increase in MAM rainfall and decrease in SON (Figure 5). Such variations might be attributable to a number of factors including variations in the micro-climatic conditions but also to the methods of analysis. The use of moving averages tries to smoothen out the variations between individual years which might be missing in other approaches. Although the seasonal rainfall for both SON and MAM revealed increase and decrease in amount over time respectively, the test statistics was not significant at (p = 0.05). It is important noting that despite the insignificance as a result of statistical analysis, there is generally a mild decline or increase in amount of seasonal rainfall over time around the study area.

Despite the linear trend analysis could not reveal a statistical significance in decline or increase in amount of seasonal and annual rainfall, historical records from elder pastoralist (who have stayed in the area for over 30 years) have reported shifts in occurrence of rainfall and its distribution over the year. It is also been revealed that there is increasing rise in temperature as compared to the past. For example, the findings from household survey in Tanzania revealed that 96.7% of the agro-pastoralist respondents had stated that the rainfall is declining compared with the past periods and that the temperature has increased [13] . It is also been accentuated that there is presently inconsistency in rainfall occurrence. Majority claim that such variability in climate associated with shifts in rainfall patterns in terms of quantity and frequency over the past three decades have had direct consequences on their livelihoods because it has changed the grazing land.

Using a case in the Uganda study area, analysis for the number of rain-days and the average historical monthly rainfall revealed the concerns posed by the agro-pastoralist. Figure 6(a) and Figure 6(b) present the number of season rain-days over the time, while Figures 7(a)-(c) present a comparison in average historical monthly rainfall for the different periods. The results (Figure 6(a) and Figure 6(b)) revealed that the number of rain-days appear to be declining with time for all seasons. It is also clear that there is a shift in occurrence of rainfall (Figures 7(a)-(c)). The shift is either indicating the increase in rainfall amount in recent times as compared to the past (upward arrow) or a change in rainfall month (left arrow). The pattern of change is very consistent in the three countries, with a shift of one month in recent time (i.e. 1990-2009).

The shift analysis indicate an increase in amount of rainfall (upward arrow) for all stations considered which contradicts the perceived decrease in rainfall amount and rain-days as revealed by both household survey and rain-days analysis using the Uganda case. Such contradiction could be a result of increase in rainfall intensity for any single event of rainfall concentrated in a very narrow time band. According to [14] [15] , climate predictions along Lake Victoria region indicate increase in temperature associated with increase in rainfall event in very isolated areas. Furthermore, climate change scenarios using multiple general circulation models show increases in average mean temperature. In Tanzania, the predictions show that the mean daily temperature will rise by 3˚C - 5˚C throughout the country and the mean annual temperature will rise by 2˚C - 4˚C. Further predictions show that areas with bimodal rainfall pattern will experience increased rainfall of 5% - 45% and those with unimodal rainfall pattern will experience decreased rainfall of 5% - 15% [14] [15] .

Temperature analyses are consistent with climate prediction models as revealed in Figure 8 and Figure 9(a), Figure 9(b). The results show a consistent increase in temperature across the study area in the Lake Victoria Basin. The only exception is for Nyagatare area in Rwanda that indicate very limited variations unlike those found in the other two countries (Figure 9(a), Figure 9(b)). According to [16] extreme temperature events constitute climate variability and change; hence the observed variations in climate might be an artefact of temperature rise in the area.

The consequence of consistent conversion of grazing areas in the LVB into other land use types in the past twenty to thirty years is reduction in pasture availability, which is the cheapest and major feed source for ruminant livestock production. Although cattle populations have been on the increase in the same period, it has been

largely attributed to the fact that the carrying capacity of available grazing areas had not been attained. The current stocking rates in the LVB reveal that the rangelands are greatly overstocked and overgrazed with degradation leading to bare areas which are already evidenced. Climate variability coupled with a decrease in grazing resources are driving unprecedented forage scarcity which is a major limiting factor to cattle production.

With crop cultivation being a major land use type overtaking grazing lands, the incorporation of crop residues into ruminant feeding areas is a feasible way out to curtail rangeland degradation and increase beef production. However, despite the immense conversion of traditional grazing land into cultivation, there is limited use of crop residues in animal feeding with lack of knowledge to change farmers’ behaviour [9] . The current trend will increase pressure on the dwindling grazing resources as new additions of stock to meet livelihood requirements are being done under restrained resources.

One hectare of maize can generate about 7.76 tons dry matter of stover [17] . With animals consuming up to 2.5% DM of body weight per day of dry forages [18] stover from one hectare of maize can sustain up to three Tropical Livestock Units (TLU) for a year in areas with one cropping season and up to six TLU in areas with two cropping seasons a year provided that farmers can balance the energy and protein requirements of animals with supplements (crop residues, forage legumes, agro processing by-products or commercial supplements).

The climate variability and change analysis has revealed the changes in rainfall and temperature in the Lake Victoria Basin. [19] cautioned that, while climate change is a global phenomenon, people will be differently affected by its local impacts. Among others, change in temperature and precipitation patterns will have a profound impact on the natural resource base worldwide, and on the income and livelihoods of people that depend on these resources. This is now a reality in the Lake Victoria Basin where the resulting effects are more apparent. Discussion with agro-pastoral communities in the three study area of LVB revealed numerous impacts and consequences as a result of climate variability and change Results from the three countries’ case studies have revealed the following impacts and consequences (Table 5).

The above findings are consistent previous studies conducted elsewhere in the LVB. For example [20] revealed that livestock production is one of the production systems that are potentially vulnerable to climate variability and change. Their findings revealed significant decrease in pastures leading to low livestock production in terms of calves, milk and meat and that unpredictable rainfall, has resulted into inadequate pastures as compared to previous years. As a result livestock production is locally believed to decline. It was asserted that because of the unfavourable pasture conditions the livestock do not have sufficient feed; they are generally weak and may not breed as efficiently as would do under favourable conditions. Generally, shortage of pastures due to changing climate may result in competition for available land, causing conflict and insecurity [20] . Their study also revealed problems associated with declining water resources, increase in diseases, declining forage productivity and other impacts beyond livestock production.