The present work addresses a study about herd structure and other related reproductive parameters from data gathered in the Spanish Pig Database BDporc®.
The mean annual productivity of 29.73 PWSY indicates efficient breeding practices, aligning well with established industry standards in Spain [20]; therefore, these farms are considered to have high annual productivity [21]. While these farms demonstrate good productivity outcomes, it is essential to note that these show means of 1.33 SB piglets and pre-weaning mortality rates of 18.4% and 14.4% for piglets TB and BA, respectively, that should be improved. In this regard, these results show piglet survival rates below the minimum target suggested by Sanz-Fernández et al. [5] which are 83.2% and 88.5% for piglets TB and BA, respectively. Therefore, while these farms exhibit high productivity there is room for improvement.
In terms of prolificacy (BA) per parity, it is well known that this varies throughout the reproductive cycles of sows [22]. In this study, the 3rd and 4th parities prove to be the most productive cycles, in line with the parity curve pattern of litter size reported by Sell-Kubiak et al. [23]. A drop of almost two piglets was observed between these highly productive cycles and the one with the lowest prolificacy (parities ≥ 8).
Besides that, as in the study by Lavery et al. [24], the WCI decreases as the number of parities increases, with a decline in the number of W piglets from the 3rd parity onwards.
Sow longevity, farrowings per culled sow, replacement rate and herd distribution are all inter-related measures. This study reports similar results to those found in a study of 110 commercial breeding herds in Japan by Koketsu [25], with a mean of 4.55 farrowings at culling and a 47.3% replacement rate. Although currently a replacement rate of 40–50% is considered appropriate for maintaining a proper herd structure [26], keeping a sow on the farm for a longer time allows for a greater opportunity to recoup the initial investment [27]. According to Małopolska [28], the primary reasons for culling sows are reproductive problems, leading to an increase in replacement of sows. This, in turn, results in higher production costs and decreased profitability [17].
Furthermore, herd longevity is a concern not only from an economic and productive perspective but also from a consumer perspective of animal welfare. Hoge and Bates [29] suggest improving sow management to extend the productive life of breeding sows to improve both profitability and animal welfare. This is particularly relevant given the increasing societal awareness and concerns about animal welfare and sustainability [30]. Thus, the use of indicators such as herd longevity may be crucial for evaluating sustainability, animal welfare and management of breeding sows.
While several studies have examined models for sow herd management [31] and sow removal and culling patterns [1, 17, 32, 33], there remains a gap in understanding the impact of herd structure on farm efficiency. This study bridges that gap by classifying farms into three distinct types or models of herd structure based on the coefficient "a" of quadratic functions fitted to the percentage of sows per parity. These models exhibit distinct shapes and characteristics, significantly contributing to our understanding of sow distribution patterns. Moreover, when analysing the influence of these three types of herd structure on farms' reproductive efficiency within a year, significant differences are observed.
HS1 shows a downward concave trend, with a higher percentage of sows in the intermediate parities (3rd to 5th ). This distribution ensures that more than 90% of the sows on the farm are within the 1st to 6th farrowing, allowing them to reach their maximum reproductive potential, as these sows are the most productive [4]. Thereby, farms HS1 achieve the best productive outcomes, including a higher number of PWSY, surpassing the results of farms with other type of herd structure.
Furthermore, Buxadé Carbó et al. [34] suggested maintaining a higher percentage of sows until the 3rd or 4th parity to maximise their depreciation. This implies that the percentage difference between the 1st and 2nd, and 2nd and 3rd cycles should be minimal, resembling the HS1 defined in this study, which has a higher percentage of sows in intermediate parities. This allows for the maximisation of the number of sows in the most productive parities, achieving higher productivity and reducing the average cost per piglet.
HS2 exhibits a trend curve closer to a straight line, maintaining a steady decline in the percentage of sows, aligning with the ideal herd structure defined by Carroll [10]. This strategy aims to mitigate productivity variations attributed to unstructured herd distribution. On average, this group showed 19.3% of first-parity sows, slightly above the 17% indicated by Carroll [10] and below the 24.3% reported in the recent cohorts study by Bergman et al. [30]. Furthermore, the HS2 also aligns with the recommendations of Houška [1], who suggested that the percentage of sows from the 1st and 2nd parity should be similar to the percentage of sows from the 3rd to the 5th parity; which, in this study, represent 36.3% and 39.6% of the breeding sows census, respectively. Therefore, this census distribution is considered a highly stable herd structure over time. Additionally, farms with HS1 and HS2 also show better productivity in terms of the number of farrowings per sow per year and fewer non-productive days, (i.e., lower values of WSI and WOI) compared to HS3 farms.
On the other hand, the census distribution of HS3 farms, with an upward concave trend and a higher percentage of sows in the latest parities (6th to 8th or older) than HS1 and HS2 farms, may be attributed to unstructured herd [1], with lower productivity results than the other two types of herd structure. This may be due to the higher proportion of older sows, which are less productive [4]. In addition, this structure had a lower coefficient of determination (R2 = 0.57) compared to HS1 and HS2 (R2 = 0.76 in both cases), indicating a slightly larger average difference between predicted and observed values when compared to the other two herd structure types. This difference in the coefficient of determination may be due to a higher variability of farms within this group, including farms with highly unstructured herd distribution. However, this poorer model fit could also be due to the grouping of sows from the 8th farrowing onwards, which could bias the quadratic function, especially in these HS3 farms (≥ 8th parities representing 12% of total sows), not accurately capturing the trend between the last parities. Unfortunately, this information was not available for inclusion in the models, as the BDporc dataset used groups sows from the 8th farrowing onwards.
In addition, the linear functions of the three types of herd structure, represented alongside the quadratic functions, exhibit a less precise model fit. Hence, this confirms that a quadratic regression model better fits the reality of the farms' census structure.
These results confirm the need to consider herd structure as a relevant factor to evaluate reproductive efficiency. They demonstrate that farms with HS2, traditionally described as ideal or model herd [1, 3, 10–13, 30], do not achieve the best productivity results over the course of a year. Therefore, it is worth questioning whether it should be considered ideal in terms of productivity, even though it maintains a constant herd size between cycles. On the other hand, HS1 achieve the best productivity results and aligns with the herd structure described by De Andrés et al. [2], who defined an ideal herd structure different from that described by Carroll [10], with fewer first-parity sows and a higher percentage of sows in the most productive cycles (3rd-4th) by culling fewer sows in these early cycles and maintaining a declining herd size. However, the present study cannot confirm that HS1 is the most productive in the long term, as it has only examined the herd structure and productivity of the farms over a year.
In this regard, a proper herd structure must ensure stable productivity over time, with its potential increase as a result of prolificacy, survival rate and fertility improvements. This can potentially be achieved with HS1 and HS2, provided that an appropriate replacement and culling policy is in place. For example, Mote et al. [35] suggested that producers should aim to limit sow losses to no more than 10% per parity cycle to maintain an ideal herd. However, in some farms, it may be beneficial to increase the percentage of sows in the later parities. In this context, Rodriguez-Zas et al. [17] recommended that in situations where sow costs are high, salvage or residual values are low, and revenues per piglet are also low, the optimal parity for removal should be between 6 to 10 parities. Additionally, maintaining a herd age structure that retains mature sows allows them to reach their maximum performance [30], which depends on the management and results of each farm, and would explain why some HS3 farms can achieve good results in terms of productivity. However, it should also be considered that old sows have a higher feed consumption [24], which increases costs of production and could reduce profitability and sustainability.
On the contrary, some farms included in the HS1 group, despite having a higher number of sows in the most productive parities (from 3rd to 5th) than HS2 and HS3, have a risk of reducing their productivity in the following year if current young sows (1st and 2nd parities) do not have a low culling rate to maintain the sow census in the future 3th to 5th parities. Therefore, that structure could lead to annual variations in productivity. Considering the above, when organising a farm, it is essential to study its optimal herd structure like any other production parameters and their targets, with the aim of maintaining a consistent replacement and culling policy over time. In any case, this study did not have information on the management techniques implemented on the farms or its health status, nor on the culling rates per parity, which represents a limitation of the study, as it would have provided relevant information to better understand the elimination patterns of different types of herd structure.